DROPGoogle launches Gemini 3.8 Flash TTS and Flash-Lite TTS///DROPGemini Omni brings 1080p generation to Google Vids///BREAKClaude Opus 5.5 is here: 1M context, $4 input / $20 output///BREAKOpenAI adds GPT-6 Sol and Luna to the GPT-6 family///PAPEROpus 5.5's system card documents gains—and evaluation limits///BREAKGrok 4.7 launches at $2 input / $6 output per million tokens///DROPGrok Voice Transcribe 2.0 targets difficult real-world audio///TRENDGoogle steers new Gemini API projects to 3.8 Flash or 3.5 Flash-Lite///DROPGoogle launches Gemini 3.8 Flash TTS and Flash-Lite TTS///DROPGemini Omni brings 1080p generation to Google Vids///BREAKClaude Opus 5.5 is here: 1M context, $4 input / $20 output///BREAKOpenAI adds GPT-6 Sol and Luna to the GPT-6 family///PAPEROpus 5.5's system card documents gains—and evaluation limits///BREAKGrok 4.7 launches at $2 input / $6 output per million tokens///DROPGrok Voice Transcribe 2.0 targets difficult real-world audio///TRENDGoogle steers new Gemini API projects to 3.8 Flash or 3.5 Flash-Lite///
DARC.DEV

[ 04 — DC Model Index ]

Choose with evidence. Build with confidence.

Compare model capability, cost and current releases. Inspect original benchmark results, source coverage, domain scores and workload costs. Compare many models with one reference before making a choice.

Update delayed · last successful import · 437 models · Sources & methodology ↗

Power combines four source families: 35% Coding, 35% Reasoning and 30% Agents. Only complete fixed panels qualify for Power, Eco and Overall. Partial domain scores are labelled with coverage; they do not establish a win against a complete panel. Compare the underlying tests or inspect every formula.

Inspect the evidence

Many models. One reference.

Compare original test results, not just a league position. Green + and red − show a numerical difference for the same named benchmark, effort and harness. Equal effort names can still use different compute budgets. Unknown setups have no coloured delta.

Hy3REFERENCEClaude Fable 5.1GPT-6 AstraGPT-6.1 Terra

97 benchmark/setup rows. Up to 12 models, including the reference. “—” means missing evidence; “conflict” means no unambiguous observation. Click a score for source details. Scroll horizontally for more models.

Eco & Overall: change your workload assumptions

Only the scenario rows below change. Standard ranking: 3,000 uncached input + 1,000 output tokens, 75% quality. Reasoning tokens, tool calls, retries, cache writes and long-context surcharges are excluded. This estimates token spend, not cost per successful task. Unknown cache prices leave a cached scenario empty.

Benchmark / setupHy3ReferenceClaude Fable 5.1anthropicGPT-6 AstraopenaiGPT-6.1 Terraopenai
DC PowerFixed multi-source panel · score points—0% fixed panel · Power withheld
Inspect Hy3: weights, inputs & missing evidence

Fixed-panel coverage: 0%. Power, Eco and Overall are withheld until every panel is available. Partial domain values use only available panels and are not directly comparable with complete domain scores. ECI contributes 0%.

Coding: — · 35% of Power · 0% coverage · PARTIAL

LiveBench 2 Coding · 25 Oct 2026 · 50% of Coding = 17.5% of Power · —

Missing: LiveBench 2 Coding · 20261025: no eligible original observation

Epoch AI original MirrorCode · 50% of Coding = 17.5% of Power · —

Missing: MirrorCode: no eligible original observation

Reasoning: — · 35% of Power · 0% coverage · PARTIAL

LiveBench 2 Reasoning · 25 Oct 2026 · 40% of Reasoning = 14.0% of Power · —

Missing: LiveBench 2 Reasoning · 20261025: no eligible original observation

Epoch AI original reasoning panel · 30% of Reasoning = 10.5% of Power · —

Missing: Chess Puzzles: no eligible original observation

Missing: FrontierMath-Tiers-1-3-v2-Private: no eligible original observation

Missing: OTIS Mock AIME 2024-2025: no eligible original observation

ARC Prize · ARC-AGI-2 semi-private · 30% of Reasoning = 10.5% of Power · —

Missing: ARC-AGI-2 · semi-private: no eligible original observation

Agents: — · 30% of Power · 0% coverage · PARTIAL

LiveBench 2 Long Agentic · 25 Oct 2026 · 50% of Agents = 15.0% of Power · —

Missing: LiveBench 2 Long Agentic · 20261025: no eligible original observation

Terminal-Bench 4.0 · published model + agent · 50% of Agents = 15.0% of Power · —

Missing: Terminal-Bench 4.0: no eligible original observation

Power — · incomplete fixed panel. Missing scores are not zero.

80.9100% fixed panel · 4 source familiesΔ unavailable
Inspect Claude Fable 5.1: weights, inputs & missing evidence

Fixed-panel coverage: 100%. All ranked models use this complete panel. ECI contributes 0%.

Coding: 85.5 · 35% of Power · 100% coverage

LiveBench 2 Coding · 25 Oct 2026 · 50% of Coding = 17.5% of Power · 97.7

LiveBench 2 Coding · 20261025: 97.70% ↗claude-fable-5.1 · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Epoch AI original MirrorCode · 50% of Coding = 17.5% of Power · 73.3

MirrorCode: 73.30% ↗claude-fable-5-1_high · Epoch Inspect · Epoch AI evaluation

Reasoning: 87.5 · 35% of Power · 100% coverage

LiveBench 2 Reasoning · 25 Oct 2026 · 40% of Reasoning = 14.0% of Power · 91.7

LiveBench 2 Reasoning · 20261025: 91.70% ↗claude-fable-5.1 · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Epoch AI original reasoning panel · 30% of Reasoning = 10.5% of Power · 79.4

Chess Puzzles: 48.00% ↗claude-fable-5-1_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

FrontierMath-Tiers-1-3-v2-Private: 90.18% ↗claude-fable-5-1_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

OTIS Mock AIME 2024-2025: 100.00% ↗claude-fable-5-1_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

ARC Prize · ARC-AGI-2 semi-private · 30% of Reasoning = 10.5% of Power · 90.0

ARC-AGI-2 · semi-private: 90.00% ↗Claude Fable 5.1 (Max) · ARC Prize leaderboard baseline · ARC Prize official leaderboard

Agents: 67.9 · 30% of Power · 100% coverage

LiveBench 2 Long Agentic · 25 Oct 2026 · 50% of Agents = 15.0% of Power · 77.9

LiveBench 2 Long Agentic · 20261025: 77.90% ↗claude-fable-5.1 · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Terminal-Bench 4.0 · published model + agent · 50% of Agents = 15.0% of Power · 57.9

Terminal-Bench 4.0: 57.88% ↗Claude Fable 5.1 · Claude Code · max · Terminal-Bench 4.0 · Claude Code · Terminal-Bench public leaderboard

Power = 0.35 × 85.5 + 0.35 × 87.5 + 0.30 × 67.9 = 80.9. Full precision is used before final rounding.

77.2100% fixed panel · 4 source familiesΔ unavailable
Inspect GPT-6 Astra: weights, inputs & missing evidence

Fixed-panel coverage: 100%. All ranked models use this complete panel. ECI contributes 0%.

Coding: 70.2 · 35% of Power · 100% coverage

LiveBench 2 Coding · 25 Oct 2026 · 50% of Coding = 17.5% of Power · 93.7

LiveBench 2 Coding · 20261025: 93.70% ↗gpt-6-astra · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Epoch AI original MirrorCode · 50% of Coding = 17.5% of Power · 46.7

MirrorCode: 46.70% ↗gpt-6-astra_high · Epoch Inspect · Epoch AI evaluation

Reasoning: 92.3 · 35% of Power · 100% coverage

LiveBench 2 Reasoning · 25 Oct 2026 · 40% of Reasoning = 14.0% of Power · 92.7

LiveBench 2 Reasoning · 20261025: 92.70% ↗gpt-6-astra · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Epoch AI original reasoning panel · 30% of Reasoning = 10.5% of Power · 89.2

Chess Puzzles: 74.00% ↗gpt-6-astra_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

FrontierMath-Tiers-1-3-v2-Private: 93.68% ↗gpt-6-astra_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

OTIS Mock AIME 2024-2025: 100.00% ↗gpt-6-astra_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

ARC Prize · ARC-AGI-2 semi-private · 30% of Reasoning = 10.5% of Power · 95.0

ARC-AGI-2 · semi-private: 95.00% ↗GPT-6 Astra (Max) · ARC Prize leaderboard baseline · ARC Prize official leaderboard

Agents: 67.8 · 30% of Power · 100% coverage

LiveBench 2 Long Agentic · 25 Oct 2026 · 50% of Agents = 15.0% of Power · 77.5

LiveBench 2 Long Agentic · 20261025: 77.50% ↗gpt-6-astra · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Terminal-Bench 4.0 · published model + agent · 50% of Agents = 15.0% of Power · 58.2

Terminal-Bench 4.0: 58.18% ↗GPT-6 Astra · Codex · max · Terminal-Bench 4.0 · Codex · Terminal-Bench public leaderboard

Power = 0.35 × 70.2 + 0.35 × 92.3 + 0.30 × 67.8 = 77.2. Full precision is used before final rounding.

—46.5% fixed panel · Power withheldΔ unavailable
Inspect GPT-6.1 Terra: weights, inputs & missing evidence

Fixed-panel coverage: 46.5%. Power, Eco and Overall are withheld until every panel is available. Partial domain values use only available panels and are not directly comparable with complete domain scores. ECI contributes 0%.

Coding: 94.3 · 35% of Power · 50% coverage · PARTIAL

LiveBench 2 Coding · 25 Oct 2026 · 50% of Coding = 17.5% of Power · 94.3

LiveBench 2 Coding · 20261025: 94.30% ↗gpt-6.1-terra · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Epoch AI original MirrorCode · 50% of Coding = 17.5% of Power · —

Missing: MirrorCode: no eligible original observation

Reasoning: 88.7 · 35% of Power · 40% coverage · PARTIAL

LiveBench 2 Reasoning · 25 Oct 2026 · 40% of Reasoning = 14.0% of Power · 88.7

LiveBench 2 Reasoning · 20261025: 88.70% ↗gpt-6.1-terra · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Epoch AI original reasoning panel · 30% of Reasoning = 10.5% of Power · —

FrontierMath-Tiers-1-3-v2-Private: 90.88% ↗gpt-6.1-terra_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

OTIS Mock AIME 2024-2025: 100.00% ↗gpt-6.1-terra_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

Missing: Chess Puzzles: no eligible original observation

ARC Prize · ARC-AGI-2 semi-private · 30% of Reasoning = 10.5% of Power · —

Missing: ARC-AGI-2 · semi-private: no eligible original observation

Agents: 48.6 · 30% of Power · 50% coverage · PARTIAL

LiveBench 2 Long Agentic · 25 Oct 2026 · 50% of Agents = 15.0% of Power · 48.6

LiveBench 2 Long Agentic · 20261025: 48.60% ↗gpt-6.1-terra · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Terminal-Bench 4.0 · published model + agent · 50% of Agents = 15.0% of Power · —

Missing: Terminal-Bench 4.0: no eligible original observation

Power — · incomplete fixed panel. Missing scores are not zero.

DC CodingPartial panels are not directly comparable—0% panel · PARTIAL85.5100% panelIncomplete panel · no aggregate delta70.2100% panelIncomplete panel · no aggregate delta94.350% panel · PARTIALIncomplete panel · no aggregate delta
DC ReasoningPartial panels are not directly comparable—0% panel · PARTIAL87.5100% panelIncomplete panel · no aggregate delta92.3100% panelIncomplete panel · no aggregate delta88.740% panel · PARTIALIncomplete panel · no aggregate delta
DC AgentsPartial panels are not directly comparable—0% panel · PARTIAL67.9100% panelIncomplete panel · no aggregate delta67.8100% panelIncomplete panel · no aggregate delta48.650% panel · PARTIALIncomplete panel · no aggregate delta
Scenario token cost (USD)Lower is cheaper$0.000578$0.08$0.08$0.008
Scenario DC ecoConditional estimate; not a measured outcome—9.08.6—
Scenario DC overallConditional estimate; not a measured outcome—62.960.1—
Evidence coverage0 ECI tests · 0 raw benchmarks
0 Epoch runs · 0 publisher rows · 0 external rows
22 ECI tests · 34 raw benchmarks
13 Epoch runs · 10 publisher rows · 43 external rows
24 ECI tests · 36 raw benchmarks
20 Epoch runs · 11 publisher rows · 46 external rows
10 ECI tests · 17 raw benchmarks
11 Epoch runs · 0 publisher rows · 8 external rows
ARC-AGI-2 · semi-privateReasoning · max
ARC Prize leaderboard baseline
ARC Prize official
—
90.00%Δ unavailable

Claude Fable 5.1 (Max)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $4.492349 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

95.00%Δ unavailable

GPT-6 Astra (Max)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $1.119468 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

—
ARC-AGI-2 · semi-privateReasoning · xhigh
ARC Prize leaderboard baseline
ARC Prize official
—
90.00%Δ unavailable

Claude Fable 5.1 (XHigh)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $3.120217 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

93.33%Δ unavailable

GPT-6 Astra (XHigh)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $0.828686 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

—
ARC-AGI-2 · semi-privateReasoning · high
ARC Prize leaderboard baseline
ARC Prize official
—
88.75%Δ unavailable

Claude Fable 5.1 (High)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $1.667306 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

92.08%Δ unavailable

GPT-6 Astra (High)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $0.668079 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

—
ARC-AGI-2 · semi-privateReasoning · medium
ARC Prize leaderboard baseline
ARC Prize official
—
86.25%Δ unavailable

Claude Fable 5.1 (Medium)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $1.223941 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

92.08%Δ unavailable

GPT-6 Astra (Medium)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $0.480093 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

—
ARC-AGI-2 · semi-privateReasoning · low
ARC Prize leaderboard baseline
ARC Prize official
—
78.33%Δ unavailable

Claude Fable 5.1 (Low)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $0.952327 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

85.42%Δ unavailable

GPT-6 Astra (Low)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $0.416242 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

—
ARC-AGI-2 · semi-privateReasoning · none
ARC Prize leaderboard baseline
ARC Prize official
——
59.58%Δ unavailable

GPT-6 Astra (None)

ARC Prize official leaderboard · benchmark-publisher

Run: Date not supplied · evaluations.json · v2_Semi_Private

Reported benchmark cost: $0.369674 per task. Different from the token-workload estimate.

Semi-private split only. Provider-adapter and custom-scaffold families are kept separate. Identical effort names do not imply identical compute budgets.

Inspect source ↗

Public leaderboard facts · ARC Prize

—
ARC-AGI-2Reasoning · max
Not reported
Epoch bundle · settings unverified
—
90.00%Setup unverified · no delta

claude-fable-5-1_max

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

95.00%Setup unverified · no delta

gpt-6-astra_max

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGI-2Reasoning · xhigh
Not reported
Epoch bundle · settings unverified
—
90.00%Setup unverified · no delta

claude-fable-5-1_xhigh

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

93.33%Setup unverified · no delta

gpt-6-astra_xhigh

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGI-2Reasoning · high
Not reported
Epoch bundle · settings unverified
—
88.75%Setup unverified · no delta

claude-fable-5-1_high

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

92.08%Setup unverified · no delta

gpt-6-astra_high

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGI-2Reasoning · medium
Not reported
Epoch bundle · settings unverified
—
86.25%Setup unverified · no delta

claude-fable-5-1_medium

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

92.08%Setup unverified · no delta

gpt-6-astra_medium

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGI-2Reasoning · low
Not reported
Epoch bundle · settings unverified
—
78.33%Setup unverified · no delta

claude-fable-5-1_low

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

85.42%Setup unverified · no delta

gpt-6-astra_low

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGI-2Reasoning · none
Not reported
Epoch bundle · settings unverified
——
59.58%Setup unverified · no delta

gpt-6-astra_none

External result; evaluator not specified · external

Run: Date not supplied · arc_agi_2_external.csv

Inspect source ↗

CC BY 4.0

—
GPQA diamondKnowledge · max
Epoch Inspect
Epoch bundle
——
95.77%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · gpqa_diamond.csv

Reported standard error: ±1.37 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

94.32%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-10-26 · gpqa_diamond.csv

Reported standard error: ±1.49 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

MirrorCodeCoding · high
Epoch Inspect
Epoch bundle
—
73.30%Δ unavailable

claude-fable-5-1_high

Epoch AI evaluation · epoch-run

Run: 2026-09-10 · mirrorcode.csv

Reported standard error: ±9.17 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

46.70%Δ unavailable

gpt-6-astra_high

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · mirrorcode.csv

Reported standard error: ±11.86 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
FrontierSWECoding · max
proximus
Epoch bundle
—
56.29%Δ unavailable

claude-fable-5-1_max

https://www.frontierswe.com/ · external

Run: Date not supplied · frontierswe_external.csv

FrontierSWE V2 (mean@5 across 34 tasks, proximus harness, max reasoning effort). Site footnote: Claude Opus 5 used as fallback for tasks blocked by content filters.

Inspect source ↗

CC BY 4.0

65.51%Δ unavailable

gpt-6-astra_max

https://www.frontierswe.com/ · external

Run: Date not supplied · frontierswe_external.csv

Inspect source ↗

CC BY 4.0

—
APEX-AgentsAgents · unknown
Not reported
Epoch bundle · settings unverified
—
68.60%Setup unverified · no delta

claude-fable-5-1_unknown

External result; evaluator not specified · external

Run: Date not supplied · apex_agents_external.csv

Inspect source ↗

CC BY 4.0

64.70%Setup unverified · no delta

gpt-6-astra_unknown

External result; evaluator not specified · external

Run: Date not supplied · apex_agents_external.csv

Inspect source ↗

CC BY 4.0

—
APEX-AgentsAgents · max
Not reported
Epoch bundle · settings unverified
———
68.30%Setup unverified · no delta

gpt-6.1-terra_max

External result; evaluator not specified · external

Run: Date not supplied · apex_agents_external.csv

Inspect source ↗

CC BY 4.0

APEX-AgentsAgents · high
Not reported
Epoch bundle · settings unverified
—
59.70%Setup unverified · no delta

claude-fable-5-1_high

External result; evaluator not specified · external

Run: Date not supplied · apex_agents_external.csv

Inspect source ↗

CC BY 4.0

——
SimpleQA VerifiedKnowledge · max
Epoch Inspect
Epoch bundle
—
70.80%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-09-01 · simpleqa_verified.csv

Reported standard error: ±1.44 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

75.60%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · simpleqa_verified.csv

Reported standard error: ±1.36 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

57.30%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-10-26 · simpleqa_verified.csv

Reported standard error: ±1.56 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

HLEKnowledge · unknown
Not reported
Epoch bundle · settings unverified
——
54.80%Setup unverified · no delta

gpt-6-astra_unknown

External result; evaluator not specified · external

Run: Date not supplied · hle_external.csv

Reported standard error: ±1.94 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
HLEKnowledge · xhigh
Not reported
Epoch bundle · settings unverified
—
46.50%Setup unverified · no delta

claude-fable-5-1_xhigh

External result; evaluator not specified · external

Run: Date not supplied · hle_external.csv

Reported standard error: ±2.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

——
ARC-AGIReasoning · max
Not reported
Epoch bundle · settings unverified
—
97.50%Setup unverified · no delta

claude-fable-5-1_max

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

97.50%Setup unverified · no delta

gpt-6-astra_max

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGIReasoning · xhigh
Not reported
Epoch bundle · settings unverified
—
96.50%Setup unverified · no delta

claude-fable-5-1_xhigh

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

98.50%Setup unverified · no delta

gpt-6-astra_xhigh

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGIReasoning · high
Not reported
Epoch bundle · settings unverified
—
96.00%Setup unverified · no delta

claude-fable-5-1_high

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

98.50%Setup unverified · no delta

gpt-6-astra_high

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGIReasoning · medium
Not reported
Epoch bundle · settings unverified
—
94.50%Setup unverified · no delta

claude-fable-5-1_medium

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

97.50%Setup unverified · no delta

gpt-6-astra_medium

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGIReasoning · low
Not reported
Epoch bundle · settings unverified
—
90.00%Setup unverified · no delta

claude-fable-5-1_low

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

96.50%Setup unverified · no delta

gpt-6-astra_low

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

—
ARC-AGIReasoning · none
Not reported
Epoch bundle · settings unverified
——
86.00%Setup unverified · no delta

gpt-6-astra_none

https://arcprize.org/leaderboard · external

Run: Date not supplied · arc_agi_external.csv

Inspect source ↗

CC BY 4.0

—
BalrogAgents · max
Not reported
Epoch bundle · settings unverified
——
68.30%Setup unverified · no delta

gpt-6-astra_max

https://balrogai.com/ · external

Run: Date not supplied · balrog_external.csv

Inspect source ↗

CC BY 4.0

—
Chess PuzzlesReasoning · promax
Epoch Inspect
Epoch bundle
——
76.00%Δ unavailable

gpt-6-astra_promax

Epoch AI evaluation · epoch-run

Run: 2026-10-15 · chess_puzzles.csv

Reported standard error: ±4.29 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
Chess PuzzlesReasoning · max
Epoch Inspect
Epoch bundle
—
48.00%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-10-13 · chess_puzzles.csv

Reported standard error: ±5.02 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

74.00%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-10-13 · chess_puzzles.csv

Reported standard error: ±4.41 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
CritPtOther · max
Not reported
Epoch bundle · settings unverified
—
29.71%Setup unverified · no delta

claude-fable-5-1_max

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

31.71%Setup unverified · no delta

gpt-6-astra_max

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

28.00%Setup unverified · no delta

gpt-6.1-terra_max

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

CritPtOther · xhigh
Not reported
Epoch bundle · settings unverified
—
31.14%Setup unverified · no delta

claude-fable-5-1_xhigh

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

31.43%Setup unverified · no delta

gpt-6-astra_xhigh

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

26.57%Setup unverified · no delta

gpt-6.1-terra_xhigh

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

CritPtOther · high
Not reported
Epoch bundle · settings unverified
—
30.29%Setup unverified · no delta

claude-fable-5-1_high

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

28.86%Setup unverified · no delta

gpt-6-astra_high

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

25.71%Setup unverified · no delta

gpt-6.1-terra_high

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

CritPtOther · medium
Not reported
Epoch bundle · settings unverified
—
29.14%Setup unverified · no delta

claude-fable-5-1_medium

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

29.14%Setup unverified · no delta

gpt-6-astra_medium

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

21.43%Setup unverified · no delta

gpt-6.1-terra_medium

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

CritPtOther · low
Not reported
Epoch bundle · settings unverified
—
27.71%Setup unverified · no delta

claude-fable-5-1_low

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

26.29%Setup unverified · no delta

gpt-6-astra_low

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

15.14%Setup unverified · no delta

gpt-6.1-terra_low

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

CritPtOther · none
Not reported
Epoch bundle · settings unverified
——
19.71%Setup unverified · no delta

gpt-6-astra_none

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

4.86%Setup unverified · no delta

gpt-6.1-terra_none

External result; evaluator not specified · external

Run: Date not supplied · critpt_external.csv

Inspect source ↗

CC BY 4.0

CursorBenchOther · max
Not reported
Epoch bundle · settings unverified
—
51.80%Setup unverified · no delta

claude-fable-5-1_max

https://cursor.com/cursorbench · external

Run: Date not supplied · cursorbench_external.csv

Inspect source ↗

CC BY 4.0

——
CursorBenchOther · xhigh
Not reported
Epoch bundle · settings unverified
—
51.60%Setup unverified · no delta

claude-fable-5-1_xhigh

https://cursor.com/cursorbench · external

Run: Date not supplied · cursorbench_external.csv

Inspect source ↗

CC BY 4.0

——
CursorBenchOther · high
Not reported
Epoch bundle · settings unverified
—
49.20%Setup unverified · no delta

claude-fable-5-1_high

https://cursor.com/cursorbench · external

Run: Date not supplied · cursorbench_external.csv

Inspect source ↗

CC BY 4.0

——
CursorBenchOther · medium
Not reported
Epoch bundle · settings unverified
—
46.80%Setup unverified · no delta

claude-fable-5-1_medium

https://cursor.com/cursorbench · external

Run: Date not supplied · cursorbench_external.csv

Inspect source ↗

CC BY 4.0

——
CursorBenchOther · low
Not reported
Epoch bundle · settings unverified
—
45.10%Setup unverified · no delta

claude-fable-5-1_low

https://cursor.com/cursorbench · external

Run: Date not supplied · cursorbench_external.csv

Inspect source ↗

CC BY 4.0

——
DeepSWECoding · max
mini-swe-agent
Epoch bundle
——
73.23%Δ unavailable

gpt-6-astra_max

https://deepswe.datacurve.ai/ · external

Run: Date not supplied · deepswe_external.csv

Inspect source ↗

CC BY 4.0

—
DeepSWECoding · xhigh
mini-swe-agent
Epoch bundle
——
74.12%Δ unavailable

gpt-6-astra_xhigh

https://deepswe.datacurve.ai/ · external

Run: Date not supplied · deepswe_external.csv

Inspect source ↗

CC BY 4.0

—
DeepSWECoding · high
mini-swe-agent
Epoch bundle
——
73.23%Δ unavailable

gpt-6-astra_high

https://deepswe.datacurve.ai/ · external

Run: Date not supplied · deepswe_external.csv

Inspect source ↗

CC BY 4.0

—
DeepSWECoding · medium
mini-swe-agent
Epoch bundle
——
72.79%Δ unavailable

gpt-6-astra_medium

https://deepswe.datacurve.ai/ · external

Run: Date not supplied · deepswe_external.csv

Inspect source ↗

CC BY 4.0

—
DeepSWECoding · low
mini-swe-agent
Epoch bundle
——
67.04%Δ unavailable

gpt-6-astra_low

https://deepswe.datacurve.ai/ · external

Run: Date not supplied · deepswe_external.csv

Inspect source ↗

CC BY 4.0

—
DTBenchReasoning · unknown
Not reported
Epoch bundle · settings unverified
—
97.60%Setup unverified · no delta

claude-fable-5-1_unknown

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

97.07%Setup unverified · no delta

gpt-6-astra_unknown

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

—
DTBenchReasoning · xhigh
Not reported
Epoch bundle · settings unverified
—
97.60%Setup unverified · no delta

claude-fable-5-1_xhigh

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

97.33%Setup unverified · no delta

gpt-6-astra_xhigh

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

—
DTBenchReasoning · high
Not reported
Epoch bundle · settings unverified
—
96.53%Setup unverified · no delta

claude-fable-5-1_high

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

96.53%Setup unverified · no delta

gpt-6-astra_high

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

—
DTBenchReasoning · medium
Not reported
Epoch bundle · settings unverified
—
97.07%Setup unverified · no delta

claude-fable-5-1_medium

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

97.07%Setup unverified · no delta

gpt-6-astra_medium

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

—
DTBenchReasoning · low
Not reported
Epoch bundle · settings unverified
—
94.13%Setup unverified · no delta

claude-fable-5-1_low

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

94.93%Setup unverified · no delta

gpt-6-astra_low

https://conceptualreasoning.ai/dtbench · external

Run: Date not supplied · dtbench_external.csv

Inspect source ↗

CC BY 4.0

—
EBR-benchOther · max
Epoch Inspect
Epoch bundle
—
57.14%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-09-08 · ebr_bench.csv

Reported standard error: ±6.02 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

76.19%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · ebr_bench.csv

Reported standard error: ±8.38 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

78.10%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-10-26 · ebr_bench.csv

Reported standard error: ±5.13 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

FrontierCodeCoding · max
codex
Epoch bundle
——
53.26%Δ unavailable

gpt-6-astra_max

https://cognition.com/frontiercode · external

Run: Date not supplied · frontiercode_external.csv

Inspect source ↗

CC BY 4.0

—
FrontierCodeCoding · medium
claude-code
Epoch bundle
—
50.91%Δ unavailable

claude-fable-5-1_medium

https://cognition.com/frontiercode · external

Run: Date not supplied · frontiercode_external.csv

Inspect source ↗

CC BY 4.0

——
FrontierMath-ErdosReasoning · max
Epoch Inspect
Epoch bundle
—
0.00%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-09-01 · frontiermath_erdos.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

2.94%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-08-28 · frontiermath_erdos.csv

Reported standard error: ±2.06 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
FrontierMath-Tier-4-v2-PrivateReasoning · max
Epoch Inspect
Epoch bundle
—
87.80%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-09-01 · frontiermath_tier_4_v2.csv

Reported standard error: ±5.17 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

97.60%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · frontiermath_tier_4_v2.csv

Reported standard error: ±2.40 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

92.68%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-10-26 · frontiermath_tier_4_v2.csv

Reported standard error: ±4.12 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

FrontierMath-Tier-4-v2-PrivateReasoning · xhigh
Epoch Inspect
Epoch bundle
——
97.60%Δ unavailable

gpt-6-astra_xhigh

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · frontiermath_tier_4_v2.csv

Reported standard error: ±2.40 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
FrontierMath-Tier-4-v2-PrivateReasoning · high
Epoch Inspect
Epoch bundle
——
97.60%Δ unavailable

gpt-6-astra_high

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · frontiermath_tier_4_v2.csv

Reported standard error: ±2.40 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
FrontierMath-Tier-4-v2-PrivateReasoning · medium
Epoch Inspect
Epoch bundle
——
97.56%Δ unavailable

gpt-6-astra_medium

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · frontiermath_tier_4_v2.csv

Reported standard error: ±2.44 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
FrontierMath-Tier-4-v2-PrivateReasoning · low
Epoch Inspect
Epoch bundle
——
87.80%Δ unavailable

gpt-6-astra_low

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · frontiermath_tier_4_v2.csv

Reported standard error: ±5.17 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
FrontierMath-Tier-4-v2-PrivateReasoning · none
Epoch Inspect
Epoch bundle
——
82.93%Δ unavailable

gpt-6-astra_none

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · frontiermath_tier_4_v2.csv

Reported standard error: ±5.95 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
FrontierMath-Tiers-1-3-v2-PrivateReasoning · max
Epoch Inspect
Epoch bundle
—
90.18%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-09-01 · frontiermath_tiers_1_3_v2.csv

Reported standard error: ±1.77 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

93.68%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · frontiermath_tiers_1_3_v2.csv

Reported standard error: ±1.44 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

90.88%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-10-26 · frontiermath_tiers_1_3_v2.csv

Reported standard error: ±1.71 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

Furniture AssemblyReasoning · max
Epoch Inspect
Epoch bundle
—
70.00%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-09-10 · furniture_assembly.csv

Reported standard error: ±5.72 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

80.00%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-09-10 · furniture_assembly.csv

Reported standard error: ±4.93 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

79.17%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-10-26 · furniture_assembly.csv

Reported standard error: ±5.08 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

GDP.pdfOther · max
Not reported
Epoch bundle · settings unverified
—
27.60%Setup unverified · no delta

claude-fable-5-1_max

https://surgehq.ai/benchmarks/gdp-pdf · external

Run: Date not supplied · gdp_pdf_external.csv

Inspect source ↗

CC BY 4.0

34.20%Setup unverified · no delta

gpt-6-astra_max

https://surgehq.ai/benchmarks/gdp-pdf · external

Run: Date not supplied · gdp_pdf_external.csv

Inspect source ↗

CC BY 4.0

—
GDP.pdfOther · high
Not reported
Epoch bundle · settings unverified
—
29.60%Setup unverified · no delta

claude-fable-5-1_high

https://surgehq.ai/benchmarks/gdp-pdf · external

Run: Date not supplied · gdp_pdf_external.csv

Inspect source ↗

CC BY 4.0

——
GSO-BenchCoding · unknown
Not reported
Epoch bundle · settings unverified
—
88.24%Setup unverified · no delta

claude-fable-5-1_unknown

https://gso-bench.github.io/index.html · external

Run: Date not supplied · gso_external.csv

Inspect source ↗

CC BY 4.0

79.41%Setup unverified · no delta

gpt-6-astra_unknown

https://gso-bench.github.io/index.html · external

Run: Date not supplied · gso_external.csv

Inspect source ↗

CC BY 4.0

—
LiveBench 2 Coding · 20261025Coding · max
LiveBench 2 · 20261025
LiveBench 2
—
97.70%Δ unavailable

claude-fable-5.1 · max · LiveBench 2 20261025

LiveBench · independent-run

Run: Date not supplied · 20261025/table.csv

Benchmark edition 25 October 2026; individual run date not supplied. One run per model; mean task score. Failed/unmeasurable tasks score zero in the source. Source effort metadata is preserved.

Inspect source ↗

Attributed public leaderboard score facts · LiveBench

93.70%Δ unavailable

gpt-6-astra · max · LiveBench 2 20261025

LiveBench · independent-run

Run: Date not supplied · 20261025/table.csv

Benchmark edition 25 October 2026; individual run date not supplied. One run per model; mean task score. Failed/unmeasurable tasks score zero in the source. Source effort metadata is preserved.

Inspect source ↗

Attributed public leaderboard score facts · LiveBench

94.30%Δ unavailable

gpt-6.1-terra · max · LiveBench 2 20261025

LiveBench · independent-run

Run: Date not supplied · 20261025/table.csv

Benchmark edition 25 October 2026; individual run date not supplied. One run per model; mean task score. Failed/unmeasurable tasks score zero in the source. Source effort metadata is preserved.

Inspect source ↗

Attributed public leaderboard score facts · LiveBench

LiveBench 2 Long Agentic · 20261025Agents · max
LiveBench 2 · 20261025
LiveBench 2
—
77.90%Δ unavailable

claude-fable-5.1 · max · LiveBench 2 20261025

LiveBench · independent-run

Run: Date not supplied · 20261025/table.csv

Benchmark edition 25 October 2026; individual run date not supplied. One run per model; mean task score. Failed/unmeasurable tasks score zero in the source. Source effort metadata is preserved.

Inspect source ↗

Attributed public leaderboard score facts · LiveBench

77.50%Δ unavailable

gpt-6-astra · max · LiveBench 2 20261025

LiveBench · independent-run

Run: Date not supplied · 20261025/table.csv

Benchmark edition 25 October 2026; individual run date not supplied. One run per model; mean task score. Failed/unmeasurable tasks score zero in the source. Source effort metadata is preserved.

Inspect source ↗

Attributed public leaderboard score facts · LiveBench

48.60%Δ unavailable

gpt-6.1-terra · max · LiveBench 2 20261025

LiveBench · independent-run

Run: Date not supplied · 20261025/table.csv

Benchmark edition 25 October 2026; individual run date not supplied. One run per model; mean task score. Failed/unmeasurable tasks score zero in the source. Source effort metadata is preserved.

Inspect source ↗

Attributed public leaderboard score facts · LiveBench

LiveBench 2 Reasoning · 20261025Reasoning · max
LiveBench 2 · 20261025
LiveBench 2
—
91.70%Δ unavailable

claude-fable-5.1 · max · LiveBench 2 20261025

LiveBench · independent-run

Run: Date not supplied · 20261025/table.csv

Benchmark edition 25 October 2026; individual run date not supplied. One run per model; mean task score. Failed/unmeasurable tasks score zero in the source. Source effort metadata is preserved.

Inspect source ↗

Attributed public leaderboard score facts · LiveBench

92.70%Δ unavailable

gpt-6-astra · max · LiveBench 2 20261025

LiveBench · independent-run

Run: Date not supplied · 20261025/table.csv

Benchmark edition 25 October 2026; individual run date not supplied. One run per model; mean task score. Failed/unmeasurable tasks score zero in the source. Source effort metadata is preserved.

Inspect source ↗

Attributed public leaderboard score facts · LiveBench

88.70%Δ unavailable

gpt-6.1-terra · max · LiveBench 2 20261025

LiveBench · independent-run

Run: Date not supplied · 20261025/table.csv

Benchmark edition 25 October 2026; individual run date not supplied. One run per model; mean task score. Failed/unmeasurable tasks score zero in the source. Source effort metadata is preserved.

Inspect source ↗

Attributed public leaderboard score facts · LiveBench

LMCAReasoning · unknown
Not reported
Epoch bundle · settings unverified
—
65.46%Setup unverified · no delta

claude-fable-5-1_unknown

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

64.11%Setup unverified · no delta

gpt-6-astra_unknown

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

—
LMCAReasoning · xhigh
Not reported
Epoch bundle · settings unverified
—
65.06%Setup unverified · no delta

claude-fable-5-1_xhigh

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

64.37%Setup unverified · no delta

gpt-6-astra_xhigh

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

—
LMCAReasoning · high
Not reported
Epoch bundle · settings unverified
—
65.09%Setup unverified · no delta

claude-fable-5-1_high

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

63.84%Setup unverified · no delta

gpt-6-astra_high

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

—
LMCAReasoning · medium
Not reported
Epoch bundle · settings unverified
—
65.32%Setup unverified · no delta

claude-fable-5-1_medium

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

62.93%Setup unverified · no delta

gpt-6-astra_medium

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

—
LMCAReasoning · low
Not reported
Epoch bundle · settings unverified
—
63.46%Setup unverified · no delta

claude-fable-5-1_low

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

62.19%Setup unverified · no delta

gpt-6-astra_low

https://conceptualreasoning.ai/lmca · external

Run: Date not supplied · lmca_external.csv

Inspect source ↗

CC BY 4.0

—
Mystery Game PuzzlesReasoning · max
Epoch Inspect
Epoch bundle
—
58.00%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-09-01 · mystery_game_puzzles.csv

Reported standard error: ±4.96 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

84.00%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · mystery_game_puzzles.csv

Reported standard error: ±3.68 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

74.00%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-10-26 · mystery_game_puzzles.csv

Reported standard error: ±4.41 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

OTIS Mock AIME 2024-2025Reasoning · max
Epoch Inspect
Epoch bundle
—
100.00%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-09-01 · otis_mock_aime_2024_2025.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

100.00%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-08-30 · otis_mock_aime_2024_2025.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

100.00%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-10-26 · otis_mock_aime_2024_2025.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

ProofBenchReasoning · unknown
Not reported
Epoch bundle · settings unverified
——
99.00%Setup unverified · no delta

gpt-6-astra_unknown

https://www.vals.ai/benchmarks/proof_bench · external

Run: Date not supplied · proofbench_external.csv

Inspect source ↗

CC BY 4.0

—
ProofBenchReasoning · max
Not reported
Epoch bundle · settings unverified
—
100.00%Setup unverified · no delta

claude-fable-5-1_max

https://www.vals.ai/benchmarks/proof_bench · external

Run: Date not supplied · proofbench_external.csv

Inspect source ↗

CC BY 4.0

——
Remote Labor IndexAgents · unknown
Not reported
Epoch bundle · settings unverified
—
17.92%Setup unverified · no delta

claude-fable-5-1_unknown

External result; evaluator not specified · external

Run: Date not supplied · rli_external.csv

Inspect source ↗

CC BY 4.0

20.83%Setup unverified · no delta

gpt-6-astra_unknown

External result; evaluator not specified · external

Run: Date not supplied · rli_external.csv

Inspect source ↗

CC BY 4.0

—
SpireClimbOther · max
Epoch Inspect
Epoch bundle
—
2.00%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-11-05 · spireclimb.csv

Reported standard error: ±1.14 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

34.00%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-11-04 · spireclimb.csv

Reported standard error: ±4.13 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

22.33%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-11-04 · spireclimb.csv

Reported standard error: ±3.39 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

SpireClimb (Slay the Spire 1, Ascension 20 Heart)Other · max
Epoch Inspect
Epoch bundle
—
0.00%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-11-05 · spireclimb_sts1_a20_heart.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

33.33%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-11-04 · spireclimb_sts1_a20_heart.csv

Reported standard error: ±6.14 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

10.00%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-11-04 · spireclimb_sts1_a20_heart.csv

Reported standard error: ±3.91 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

SpireClimb (Slay the Spire 2, Ascension 10)Other · max
Epoch Inspect
Epoch bundle
—
4.00%Δ unavailable

claude-fable-5-1_max

Epoch AI evaluation · epoch-run

Run: 2026-09-25 · spireclimb_sts2_a10.csv

Reported standard error: ±2.28 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

34.67%Δ unavailable

gpt-6-astra_max

Epoch AI evaluation · epoch-run

Run: 2026-09-22 · spireclimb_sts2_a10.csv

Reported standard error: ±5.53 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

34.67%Δ unavailable

gpt-6.1-terra_max

Epoch AI evaluation · epoch-run

Run: 2026-11-03 · spireclimb_sts2_a10.csv

Reported standard error: ±5.53 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

Terminal-Bench 4.0Agents · max
Terminal-Bench 4.0 · Claude Code
Terminal-Bench 4.0
—
57.88%Δ unavailable

Claude Fable 5.1 · Claude Code · max

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-01 · 4-0-0/c741608e-c94e-417d-b7a8-e67111a0c887

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 3.76 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

——
Terminal-Bench 4.0Agents · max
Terminal-Bench 4.0 · Codex
Terminal-Bench 4.0
——
58.18%Δ unavailable

GPT-6 Astra · Codex · max

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-03 · 4-0-0/5c537be4-7fc3-449b-8bfc-ceb9061c2535

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 2.79 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

—
Terminal-Bench 4.0Agents · xhigh
Terminal-Bench 4.0 · Claude Code
Terminal-Bench 4.0
—
57.88%Δ unavailable

Claude Fable 5.1 · Claude Code · xhigh

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-01 · 4-0-0/9e1c5e11-748f-4f2e-911a-070a396d5bf8

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 3.36 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

——
Terminal-Bench 4.0Agents · xhigh
Terminal-Bench 4.0 · Codex
Terminal-Bench 4.0
——
57.88%Δ unavailable

GPT-6 Astra · Codex · xhigh

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-03 · 4-0-0/16db8ad5-84aa-4588-b660-1ce68c0d45e2

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 2.72 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

—
Terminal-Bench 4.0Agents · high
Terminal-Bench 4.0 · Claude Code
Terminal-Bench 4.0
—
54.55%Δ unavailable

Claude Fable 5.1 · Claude Code · high

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-01 · 4-0-0/4643b23c-53e8-47ae-a3bd-9abc900e9588

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 3.44 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

——
Terminal-Bench 4.0Agents · high
Terminal-Bench 4.0 · Codex
Terminal-Bench 4.0
——
57.88%Δ unavailable

GPT-6 Astra · Codex · high

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-03 · 4-0-0/3475050c-bf5e-4261-a3f6-0af5350af13f

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 2.97 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

—
Terminal-Bench 4.0Agents · medium
Terminal-Bench 4.0 · Claude Code
Terminal-Bench 4.0
—
53.94%Δ unavailable

Claude Fable 5.1 · Claude Code · medium

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-01 · 4-0-0/83e6bf7e-aeb4-4116-9f30-ca8de96603fb

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 3.39 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

——
Terminal-Bench 4.0Agents · medium
Terminal-Bench 4.0 · Codex
Terminal-Bench 4.0
——
54.24%Δ unavailable

GPT-6 Astra · Codex · medium

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-03 · 4-0-0/f3c3d5a6-6424-4acb-bfcc-615c3f79f6dd

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 2.66 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

—
Terminal-Bench 4.0Agents · low
Terminal-Bench 4.0 · Claude Code
Terminal-Bench 4.0
—
43.33%Δ unavailable

Claude Fable 5.1 · Claude Code · low

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-01 · 4-0-0/e18d02ab-0973-4d86-872f-a353c8a7f344

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 3.61 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

——
Terminal-Bench 4.0Agents · low
Terminal-Bench 4.0 · Codex
Terminal-Bench 4.0
——
50.61%Δ unavailable

GPT-6 Astra · Codex · low

Terminal-Bench public leaderboard · benchmark-publisher

Run: 2026-09-03 · 4-0-0/b3ad58f3-b311-4d4d-875b-3158cf0309d6

Published model + agent result; agents differ across models. Not a controlled model-only evaluation. Reported trials: 330. Source 95% accuracy interval half-width: 2.75 pp; this is not a DC Power interval.

Inspect source ↗

Attributed public leaderboard score facts · Terminal-Bench

—
WeirdMLCoding · promax
Not reported
Epoch bundle · settings unverified
——
93.57%Setup unverified · no delta

gpt-6-astra_promax

https://htihle.github.io/weirdml.html · external

Run: Date not supplied · weirdml_external.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
WeirdMLCoding · max
Not reported
Epoch bundle · settings unverified
—
92.90%Setup unverified · no delta

claude-fable-5-1_max

https://htihle.github.io/weirdml.html · external

Run: Date not supplied · weirdml_external.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

93.28%Setup unverified · no delta

gpt-6-astra_max

https://htihle.github.io/weirdml.html · external

Run: Date not supplied · weirdml_external.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
WeirdMLCoding · high
Not reported
Epoch bundle · settings unverified
—
92.30%Setup unverified · no delta

claude-fable-5-1_high

https://htihle.github.io/weirdml.html · external

Run: Date not supplied · weirdml_external.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

92.87%Setup unverified · no delta

gpt-6-astra_high

https://htihle.github.io/weirdml.html · external

Run: Date not supplied · weirdml_external.csv

Reported standard error: ±0.00 pp. This is not a 90% interval.

Inspect source ↗

CC BY 4.0

—
WeirdML v3Coding · xhigh
claude_code
Epoch bundle
—
43.53%Δ unavailable

claude-fable-5-1_xhigh

https://htihle.github.io/weirdml.html · external

Run: Date not supplied · weirdml_v3_external.csv

Inspect source ↗

CC BY 4.0

——
WeirdML v3Coding · xhigh
codex_cli
Epoch bundle
——
54.05%Δ unavailable

gpt-6-astra_xhigh

https://htihle.github.io/weirdml.html · external

Run: Date not supplied · weirdml_v3_external.csv

Inspect source ↗

CC BY 4.0

40.45%Δ unavailable

gpt-6.1-terra_xhigh

https://htihle.github.io/weirdml.html · external

Run: Date not supplied · weirdml_v3_external.csv

Inspect source ↗

CC BY 4.0

Numerical benchmark differences are descriptive, not significance tests. No confidence interval is available for the DC aggregate; the ECI source interval applies only to ECI. External rows can contain model-developer reports. Public benchmark results do not establish performance in your application.

Filter by release date, provider, scores and cost

Date filters include verified release dates only. Scores and costs marked — do not match a numeric filter. Domain sorts show complete panels first, followed by partial summaries. Domain filters can include partial scores; inspect coverage before comparing them. Task cost uses 3,000 input and 1,000 output tokens.

437 matching models · “—” means insufficient comparable evidence. Prices are listed catalogue rates; open a model to check the provider. Listings do not confirm account access.

ModelDC Power / coverageCodingReasoningAgentsDC EcoDC OverallInput / outputContextRelease / listingCompare
Claude Opus 5.5

anthropic · tools · Azure listed
21 ECI tests · 32 raw benchmarks

82.9100% fixed panel · 4 source families 87.6 100% panel90.2 100% panel68.8 100% panel19.7 67.1 $4 / $20

USD / 1M tokens

1M2026-09-22

Release date

Claude Fable 5.1

anthropic · tools · Azure listed
22 ECI tests · 34 raw benchmarks

80.9100% fixed panel · 4 source families 85.5 100% panel87.5 100% panel67.9 100% panel9.0 62.9 $10 / $50

USD / 1M tokens

1M2026-09-01

Release date

GPT-6 Astra

openai · tools · Azure listed
24 ECI tests · 36 raw benchmarks

77.2100% fixed panel · 4 source families 70.2 100% panel92.3 100% panel67.8 100% panel8.6 60.1 $10 / $50

USD / 1M tokens

1.05M2026-09-04

Release date

GPT-6 Luna

openai · tools · Azure listed
18 ECI tests · 27 raw benchmarks

51.7100% fixed panel · 4 source families 55.6 100% panel71.9 100% panel23.6 100% panel47.9 50.8 $0.1 / $0.5

USD / 1M tokens

1.05M2026-09-22

Release date

Grok 4.6

x-ai · tools · Azure listed
20 ECI tests · 29 raw benchmarks

56.8100% fixed panel · 4 source families 65.1 100% panel76.4 100% panel24.1 100% panel25.8 49.0 $2 / $6

USD / 1M tokens

500K2026-08-12

Release date

Aion 3.5

aion-labs · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $3 / $6

USD / 1M tokens

262K2026-09-23

Release date

Aion 3.5 Mini

aion-labs · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.7 / $1.4

USD / 1M tokens

262K2026-09-23

Release date

Aion-2.0

aion-labs · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.8 / $1.6

USD / 1M tokens

131K2025-12-21

Release date

Aion-3.0

aion-labs · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $3 / $6

USD / 1M tokens

131K2026-05-05

Release date

Aion-3.0-Mini

aion-labs · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.7 / $1.4

USD / 1M tokens

131K2026-05-14

Release date

Aion-RP 1.0 (8B)

aion-labs
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.8 / $1.6

USD / 1M tokens

33K2025-02-04

Release date

Beam 1.0 501B A23B

reflection · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.3 / $1

USD / 1M tokens

1.05M2026-10-05

Release date

Claude Fable 5

anthropic · tools · Azure listed
25 ECI tests · 33 raw benchmarks

—53.5% fixed panel · Power withheld 63.9 50% panel83.5 60% panel44.5 50% panel— — $10 / $50

USD / 1M tokens

1M2026-06-09

Release date

Claude Haiku 4.5

anthropic · tools
16 ECI tests · 17 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $1 / $5

USD / 1M tokens

200K2025-10-15

Release date

Claude Haiku 4.5 (latest)

anthropic · tools
0 ECI tests · 17 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $1 / $5

USD / 1M tokens

200K2025-10-15

Release date

Claude Haiku 5.5

anthropic · tools · Azure listed
18 ECI tests · 25 raw benchmarks

—67.5% fixed panel · Power withheld 91.4 50% panel74.2 100% panel40.0 50% panel— — $0.1 / $0.5

USD / 1M tokens

1M2026-10-07

Release date

Claude Opus 4.1 (latest)

anthropic · tools
0 ECI tests · 21 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $15 / $75

USD / 1M tokens

200K2025-08-05

Release date

Claude Opus 4.5

anthropic · tools
29 ECI tests · 34 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $5 / $25

USD / 1M tokens

200K2025-11-24

Release date

Claude Opus 4.5 (latest)

anthropic · tools
0 ECI tests · 34 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $5 / $25

USD / 1M tokens

200K2025-11-24

Release date

Claude Opus 4.6

anthropic · tools
26 ECI tests · 32 raw benchmarks

—10.5% fixed panel · Power withheld — 0% panel59.4 30% panel— 0% panel— — $5 / $25

USD / 1M tokens

1M2026-02-05

Release date

Claude Opus 4.7

anthropic · tools · Azure listed
24 ECI tests · 31 raw benchmarks

—28% fixed panel · Power withheld 31.1 50% panel66.0 30% panel— 0% panel— — $5 / $25

USD / 1M tokens

1M2026-04-16

Release date

Claude Opus 4.8

anthropic · tools · Azure listed
24 ECI tests · 32 raw benchmarks

—25.5% fixed panel · Power withheld — 0% panel76.8 30% panel23.6 50% panel— — $5 / $25

USD / 1M tokens

1M2026-05-28

Release date

Claude Opus 5

anthropic · tools · Azure listed
23 ECI tests · 31 raw benchmarks

—36% fixed panel · Power withheld — 0% panel83.1 60% panel51.8 50% panel— — $5 / $25

USD / 1M tokens

1M2026-07-24

Release date

Claude Sonnet 4.5

anthropic · tools
25 ECI tests · 31 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $3 / $15

USD / 1M tokens

200K2025-09-29

Release date

Claude Sonnet 4.5 (latest)

anthropic · tools
0 ECI tests · 31 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $3 / $15

USD / 1M tokens

1M2025-09-29

Release date

Claude Sonnet 4.6

anthropic · tools · Azure listed
20 ECI tests · 26 raw benchmarks

—10.5% fixed panel · Power withheld — 0% panel58.3 30% panel— 0% panel— — $3 / $15

USD / 1M tokens

1M2026-02-17

Release date

Claude Sonnet 5

anthropic · tools · Azure listed
16 ECI tests · 21 raw benchmarks

—25.5% fixed panel · Power withheld — 0% panel63.5 30% panel12.4 50% panel— — $2 / $10

USD / 1M tokens

1M2026-06-30

Release date

Claude Sonnet 5.5

anthropic · tools · Azure listed
18 ECI tests · 30 raw benchmarks

—82.5% fixed panel · Power withheld 94.7 50% panel86.6 100% panel62.6 100% panel— — $2 / $10

USD / 1M tokens

1M2026-09-28

Release date

Codestral (latest)

mistral · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.3 / $0.9

USD / 1M tokens

256K2024-05-29

Release date

Codestral 25.08

mistralai · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.3 / $0.9

USD / 1M tokens

256K2025-07-30

Release date

Command A

cohere · tools
0 ECI tests · 1 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $2.5 / $10

USD / 1M tokens

256K2025-03-13

Release date

Command A+

cohere · tools
0 ECI tests · 1 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.3 / $1.5

USD / 1M tokens

192K2026-05-20

Release date

Command R

cohere · tools
0 ECI tests · 4 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.15 / $0.6

USD / 1M tokens

128K2024-08-30

Release date

Command R+

cohere · tools
0 ECI tests · 4 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $2.5 / $10

USD / 1M tokens

128K2024-08-30

Release date

Command R7B

cohere · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.0375 / $0.15

USD / 1M tokens

128K2024-12-13

Release date

Crouton Alpha

stealth · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0 / $0

USD / 1M tokens

1.05M2026-11-03

Release date

Cydonia 24B V4.1

thedrummer
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.3 / $0.5

USD / 1M tokens

131K2025-09-27

Release date

Daybreak Blue

openai · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $4 / $20

USD / 1M tokens

1.05M2026-08-07

Release date

Daybreak Red

openai · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $12.5 / $75

USD / 1M tokens

400K2026-08-07

Release date

Deep Research Max Preview (Apr-21-2026)

google · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $2 / $12

USD / 1M tokens

131K2026-04-21

Release date

Deep Research Preview (Apr-21-2026)

google · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $2 / $12

USD / 1M tokens

131K2026-04-21

Release date

DeepSeek Chat

deepseek · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.32 / $0.89

USD / 1M tokens

164K2025-12-01

Release date

DeepSeek R1 0528

deepseek · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.5 / $2.15

USD / 1M tokens

164K2025-05-28

Release date

DeepSeek V3 0324

deepseek · tools
0 ECI tests · 0 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.29 / $1.14

USD / 1M tokens

164K2025-03-24

Release date

DeepSeek V3.1

deepseek · tools
5 ECI tests · 5 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.25 / $0.95

USD / 1M tokens

164K2025-08-21

Release date

DeepSeek V3.1 Terminus

deepseek · tools
0 ECI tests · 3 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.27 / $1

USD / 1M tokens

164K2025-09-22

Release date

DeepSeek V3.2

deepseek · tools
14 ECI tests · 14 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.26 / $0.38

USD / 1M tokens

164K2025-12-01

Release date

DeepSeek V3.2 Exp

deepseek · tools
8 ECI tests · 9 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.27 / $0.41

USD / 1M tokens

164K2025-09-29

Release date

DeepSeek V4 Flash

deepseek · tools · Azure listed
4 ECI tests · 5 raw benchmarks

—0% fixed panel · Power withheld — 0% panel— 0% panel— 0% panel— — $0.091 / $0.182

USD / 1M tokens

1.05M2026-04-24

Release date

DeepSeek V4 Flash 0731

deepseek · tools
15 ECI tests · 17 raw benchmarks

—21% fixed panel · Power withheld — 0% panel61.4 60% panel— 0% panel— — $0.0434 / $0.3

USD / 1M tokens

1.05M2026-07-31

Release date

Showing 50 of 437 matching models

DC Score v3 / independent source families

One fixed panel. Every weight visible.

DC Power is our weighted aggregate of raw benchmark scores from LiveBench, Epoch AI, ARC Prize and Terminal-Bench. It is not an ECI rescaling. Power = 0.35 × Coding + 0.35 × Reasoning + 0.30 × Agents. These are explicit editorial preferences, not statistically learned weights or a probability of success.

DomainPower weightSource panels within the domain
Coding35%

50% LiveBench 2 Coding · 25 Oct 2026

50% Epoch AI original MirrorCode

Reasoning35%

40% LiveBench 2 Reasoning · 25 Oct 2026

30% Epoch AI original reasoning panel

30% ARC Prize · ARC-AGI-2 semi-private

Agents30%

50% LiveBench 2 Long Agentic · 25 Oct 2026

50% Terminal-Bench 4.0 · published model + agent

Source balance. At full coverage, LiveBench contributes 46.5%, original Epoch AI runs 28%, ARC Prize 10.5% and Terminal-Bench 15%. One source does not gain weight by publishing more rows. ECI, vendor reports and external scores republished through Epoch have zero Power weight. Direct ARC and Terminal results enter once.

Epoch reasoning panel. Equal weights for original Epoch-run Chess Puzzles, FrontierMath tiers 1–3 v2 private and OTIS Mock AIME 2024–2025. All three are required. GPQA and other supplementary tests stay in the matrix; vendor-only GPQA reports do not become independent Epoch runs.

Missing evidence stays visible. Power, Eco and Overall require every fixed source panel. Domain scores can show a partial summary: available panel weights are renormalized, with coverage and missing inputs shown. A partial score is not a win against a complete score. The same complete panel is required for every ranked model.

Fixed editions and effort policy. LiveBench 2 edition 25 October 2026 and Terminal-Bench 4.0. Select known effort in this order: max, xhigh, high, medium, low, minimal, none, declared highest-supported. Never select the largest numerical result. A dated later run can replace an earlier run under the same setup; unresolved conflicts or different harnesses are missing.

Publisher independence is limited. Four separate source organizations do not imply blind or independently reproduced runs, independent funding, equal compute budgets or uncorrelated tests. Terminal-Bench measures a model with its published agent; Claude Code, Codex and other harnesses differ. Inspect the selected configurations. Public benchmark data cannot establish outcomes in your application.

No invented confidence interval. We do not convert ECI uncertainty into uncertainty for this aggregate. No statistically valid DC Power interval is available. Raw means are averaged on their original 0–100 scales; test difficulty and contamination can affect results. A small aggregate difference is descriptive, not proof of superiority.

Token cost C = (3,000 × input rate + 1,000 × output rate) / 1,000,000, uncached. DC Eco = Power / (1 + 100 × C), eligible only for complete Power ≥ 50 and known prices. At $0.01, Eco is half Power. DC Overall = 0.75 × Power + 0.25 × Eco. Use the comparison calculator for your workload and quality weight.

Cost is conditional. Prices are listed USD per million tokens. Reasoning tokens, tools, retries, cache writes, discounts and long-context surcharges are excluded. These are token-spend estimates, not measured cost per successful task. Quality benchmark effort does not establish a matching token workload.

Inspect Claude Fable 5.1: weights, inputs & missing evidence

Fixed-panel coverage: 100%. All ranked models use this complete panel. ECI contributes 0%.

Coding: 85.5 · 35% of Power · 100% coverage

LiveBench 2 Coding · 25 Oct 2026 · 50% of Coding = 17.5% of Power · 97.7

LiveBench 2 Coding · 20261025: 97.70% ↗claude-fable-5.1 · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Epoch AI original MirrorCode · 50% of Coding = 17.5% of Power · 73.3

MirrorCode: 73.30% ↗claude-fable-5-1_high · Epoch Inspect · Epoch AI evaluation

Reasoning: 87.5 · 35% of Power · 100% coverage

LiveBench 2 Reasoning · 25 Oct 2026 · 40% of Reasoning = 14.0% of Power · 91.7

LiveBench 2 Reasoning · 20261025: 91.70% ↗claude-fable-5.1 · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Epoch AI original reasoning panel · 30% of Reasoning = 10.5% of Power · 79.4

Chess Puzzles: 48.00% ↗claude-fable-5-1_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

FrontierMath-Tiers-1-3-v2-Private: 90.18% ↗claude-fable-5-1_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

OTIS Mock AIME 2024-2025: 100.00% ↗claude-fable-5-1_max · Epoch Inspect · Epoch AI evaluation · each input 33.33% of this panel

ARC Prize · ARC-AGI-2 semi-private · 30% of Reasoning = 10.5% of Power · 90.0

ARC-AGI-2 · semi-private: 90.00% ↗Claude Fable 5.1 (Max) · ARC Prize leaderboard baseline · ARC Prize official leaderboard

Agents: 67.9 · 30% of Power · 100% coverage

LiveBench 2 Long Agentic · 25 Oct 2026 · 50% of Agents = 15.0% of Power · 77.9

LiveBench 2 Long Agentic · 20261025: 77.90% ↗claude-fable-5.1 · max · LiveBench 2 20261025 · LiveBench 2 · 20261025 · LiveBench

Terminal-Bench 4.0 · published model + agent · 50% of Agents = 15.0% of Power · 57.9

Terminal-Bench 4.0: 57.88% ↗Claude Fable 5.1 · Claude Code · max · Terminal-Bench 4.0 · Claude Code · Terminal-Bench public leaderboard

Power = 0.35 × 85.5 + 0.35 × 87.5 + 0.30 × 67.9 = 80.9. Full precision is used before final rounding.

Reference token cost: $0.08 · Eco: 9.0 · Overall: 62.9

v3 replaces the previous ECI transform; scores across methodology versions are not directly comparable. Data refreshes every six hours, not on each page view. Benchmark editions are pinned for comparability; changing the test panel requires a methodology revision. Imports older than 24 hours are marked delayed. Catalogue releases continue to update independently. Azure listings do not confirm account quota or regional capacity.

OpenRouter catalogue ↗ · Public catalogue API; attributed metadata and list prices

models.dev ↗ · MIT

Epoch AI, Capabilities & benchmarking ↗ · CC BY 4.0; only original Epoch-run panel observations enter Power; ECI has zero weight

ARC Prize official leaderboard ↗ · Attributed public leaderboard facts; semi-private split; one direct Power panel, never counted through Epoch again

LiveBench 2 · 25 October 2026 ↗ · Attributed public score facts; original Coding, Reasoning and Long Agentic suites

Terminal-Bench 4.0 ↗ · Attributed public score facts; published model plus agent, not a model-only ablation

Microsoft Azure Retail Prices ↗ · Public retail meters; deployment eligibility is not confirmed

Epoch data: CC BY 4.0. Other publishers: attributed public leaderboard score facts. DARC.DEV defines the aggregate and makes no claim of source endorsement. No private account details are published.