Models · Anthropic · Out sinceReleased 1 Sep 2026

Claude Fable 5.1#4 for writing.#4 for writing, best at max effort.

Available on OpenRouterOn OpenRouterAvailable nowGenerally availableClosed (can't be downloaded)ProprietaryOur pick: Best for long, multi-step tasksPick: Best for long-running agents

Where you can use it:In apps:Claude (Team plan) · Fable 5.1

Claude Fable 5.1 is made by Anthropic. Among the models we track it ranks #4 for writing, #5 for research and analysis, #5 for coding. It's expensive to use. When an AI has to work on its own for a long time — using a computer, browsing, or handling a whole task end-to-end — this one is the most reliable.

API id claude-fable-5-1. Same model as Claude Mythos 5.1 with stricter safeguards (fallback to Opus models on cyber/bio). Default effort high (high in Claude Code, medium in Cowork/claude.ai). Cache reads $0.25/M (75% cheaper than Fable 5).

Writing & creativity: 74.9 / 100 · #4
Research & analysis: 80.1 / 100 · #5
Coding: 81.8 / 100 · #5
Price: Expensive$10 / $50
Price per 1M (blended)Blended / 1M: $20
MemoryContext: 1M
Longest answerMax output: 128K
Test resultsResults: 205 (135 independent135 indep.)
Out sinceReleased: 1 Sep 2026
Made byVendor: Anthropic
UnderstandsInputs: text, image

Scores are out of 100 for each area, compared with every model we track; “#” is its rank. Context is how much text it can read at once (1M is roughly 700,000 words). Blended price mixes the cost of what you send and what it writes back.

Thinking levels

Reasoning settings

LowMediumHighExtra highMax

Highlighted: where it did best for writing (Maximum thinking).

Highlighted: dominant setting in its writing composite.

Links

Using it through OpenRouter

On OpenRouter

OpenRouter is a service that gives access to many AI models in one place. This model is listed there as anthropic/claude-fable-5.1.

Route it as anthropic/claude-fable-5.1 at $10 in / $50 out per 1M tokens, 1M context. Listed since 1 Sep 2026.

include_reasoningmax_completion_tokensmax_tokensreasoningreasoning_effortresponse_formatstopstructured_outputstool_choicetoolsverbosity

Thinking level

Reasoning effort

How long should Claude Fable 5.1Where Claude Fable 5.1think?peaks.

Each chart is one test. Left to right, the model thinks longer; higher is a better result. The yellow ring marks where it did best. Thinking level is how long the AI thinks before it answers. Higher is slower and costs more, and it isn't always better.

Score per reasoning setting on each benchmark with ≥2 settings; ring = peak. Canonical results (independent over vendor).

Artificial Analysis Coding Agent Index

Best at MaxBest at Max

Terminal-Bench 4.0

Best at Extra highBest at Extra high

CursorBench

Best at MaxBest at Max

DeepSWE

Best at MaxBest at Max

FrontierCode

Best at Medium, worse abovePeaks at Medium

CursorBench 3.2.0

Best at MaxBest at Max

SWE Atlas - Codebase QnA

Best at Extra high, worse abovePeaks at Extra high

Terminal-Bench 2.1

Best at MaxBest at Max

SciCode

Best at MaxBest at Max

SWE-bench Pro (Anthropic internal subset)

Best at HighBest at High

Artificial Analysis Intelligence Index

Best at MaxBest at Max

Terminal-Bench-Science 0.1

Best at MaxBest at Max

AA-Briefcase v1.1

Best at MaxBest at Max

AA-LCR

Best at MaxBest at Max

AA-Omniscience Accuracy

Best at MaxBest at Max

AA-Omniscience Hallucination Rate

Best at Low, worse abovePeaks at Low

Lower is better on this test

AA-Omniscience Index

Best at MaxBest at Max

ARC-AGI-2

Best at Extra highBest at Extra high

GDPval-AA v2.1

Best at MaxBest at Max

GPQA Diamond

Best at MaxBest at Max

Harvey LAB-AA

Best at Extra high, worse abovePeaks at Extra high

Humanity's Last Exam

Best at MaxBest at Max

Humanity's Last Exam (with tools)

Best at Extra high, worse abovePeaks at Extra high

WANDR

Best at MaxBest at Max

Every test result

Every result

All the numbers,with where they came from.

Every result we've found for this model, with who measured it and a link to where we read it.

All raw rows (vendor and independent kept separate), with setting label, source, date, harness and notes.

All results for this model
						NotesNotes
AA Analyst Agent	57.5%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run; scores move in 1.25-pt steps (small task set)
AA-Briefcase	1694	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—
AA-Briefcase v1.1	1669	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—
AA-Briefcase v1.1	1678	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—
AA-Briefcase v1.1	1678	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—
AA-LCR	82.3%	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-LCR accuracy, AA-run (long multi-document reasoning)
AA-LCR	84.7%	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-LCR accuracy, AA-run (long multi-document reasoning)
AA-LCR	83.7%	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-LCR accuracy, AA-run (long multi-document reasoning)
AA-LCR	83.0%	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-LCR accuracy, AA-run (long multi-document reasoning)
AA-LCR	85.3%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-LCR accuracy, AA-run (long multi-document reasoning)
AA-Omniscience Accuracy	60.2%	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—
AA-Omniscience Accuracy	63.1%	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—
AA-Omniscience Accuracy	64.9%	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—
AA-Omniscience Accuracy	66.2%	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—
AA-Omniscience Accuracy	67.2%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—
AA-Omniscience Hallucination Rate	65.6%	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	share of non-correct answers that were wrong instead of abstaining; = 1 - AA 'omniscienceNonHallucination' field (equals breakdown.hallucinationRate on the eval page)
AA-Omniscience Hallucination Rate	69.1%	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	share of non-correct answers that were wrong instead of abstaining; = 1 - AA 'omniscienceNonHallucination' field (equals breakdown.hallucinationRate on the eval page)
AA-Omniscience Hallucination Rate	68.8%	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	share of non-correct answers that were wrong instead of abstaining; = 1 - AA 'omniscienceNonHallucination' field (equals breakdown.hallucinationRate on the eval page)
AA-Omniscience Hallucination Rate	70.5%	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	share of non-correct answers that were wrong instead of abstaining; = 1 - AA 'omniscienceNonHallucination' field (equals breakdown.hallucinationRate on the eval page)
AA-Omniscience Hallucination Rate	72.6%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	share of non-correct answers that were wrong instead of abstaining; = 1 - AA 'omniscienceNonHallucination' field (equals breakdown.hallucinationRate on the eval page)
AA-Omniscience Index	34.1	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-Omniscience Index (-100..100): correct minus incorrect, abstentions not penalised
AA-Omniscience Index	37.6	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-Omniscience Index (-100..100): correct minus incorrect, abstentions not penalised
AA-Omniscience Index	40.8	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-Omniscience Index (-100..100): correct minus incorrect, abstentions not penalised
AA-Omniscience Index	42.4	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-Omniscience Index (-100..100): correct minus incorrect, abstentions not penalised
AA-Omniscience Index	43.5	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-Omniscience Index (-100..100): correct minus incorrect, abstentions not penalised
ARC-AGI-1	97.5%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—
ARC-AGI-2	86.2%	MediumClaude Fable 5.1 (Medium)	Independent testIndependentARC Prize ↗	30 Sep 2026	—	semi-private set; $1.22/task; data https://arcprize.org/media/data/leaderboard/v2.json
ARC-AGI-2	88.8%	HighClaude Fable 5.1 (High)	Independent testIndependentARC Prize ↗	30 Sep 2026	—	semi-private set; $1.67/task; data https://arcprize.org/media/data/leaderboard/v2.json
ARC-AGI-2	90.0%	Extra highClaude Fable 5.1 (XHigh)	Independent testIndependentARC Prize ↗	30 Sep 2026	—	semi-private set; $3.12/task; data https://arcprize.org/media/data/leaderboard/v2.json
ARC-AGI-2	90.0%	MaxClaude Fable 5.1 (Max)	Independent testIndependentARC Prize ↗	30 Sep 2026	—	semi-private set; $4.49/task; data https://arcprize.org/media/data/leaderboard/v2.json
ARC-AGI-2	90.0%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—
Artificial Analysis Coding Agent Index	61.7	Extra highClaude Fable 5.1 XHigh + SWE-2 Medium	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	Devin Fusion CLI	agent Devin Fusion CLI; components: DeepSWE v1.1 63.1, SWE-Atlas-QnA 65.9, Terminal-Bench v4 56.1; avg cost $7.90/task; avg wall time 36 min/task; Devin Fusion = Fable/GPT planner + Cognition SWE-2 (medium) sidekick
Artificial Analysis Coding Agent Index	62.2	MaxFable 5.1 (max) (with fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	Claude Code	agent Claude Code; components: DeepSWE v1.1 64.3, SWE-Atlas-QnA 64.8, Terminal-Bench v4 57.6; avg cost $12.39/task; avg wall time 35 min/task; with fallback model
Artificial Analysis Intelligence Index	46.8	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA slug claude-fable-5-1-low; list price $10/50 per 1M in/out; cost to run AA Intelligence Index $2.37/task
Artificial Analysis Intelligence Index	48.9	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA slug claude-fable-5-1-medium; list price $10/50 per 1M in/out; cost to run AA Intelligence Index $2.98/task
Artificial Analysis Intelligence Index	51.2	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA slug claude-fable-5-1-high; list price $10/50 per 1M in/out; cost to run AA Intelligence Index $3.91/task
Artificial Analysis Intelligence Index	53.2	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA slug claude-fable-5-1-xhigh; list price $10/50 per 1M in/out; cost to run AA Intelligence Index $5.98/task
Artificial Analysis Intelligence Index	53.4	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA slug claude-fable-5-1; list price $10/50 per 1M in/out; cost to run AA Intelligence Index $7.63/task
Artificial Analysis output speed	47 tok/s	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	tokens/sec (median output speed, first-party API); TTFT 4.5s; list price $10/50 per 1M in/out
Artificial Analysis output speed	49 tok/s	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	tokens/sec (median output speed, first-party API); TTFT 8.1s; list price $10/50 per 1M in/out
Artificial Analysis output speed	51 tok/s	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	tokens/sec (median output speed, first-party API); TTFT 29.7s; list price $10/50 per 1M in/out
Artificial Analysis output speed	59 tok/s	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	tokens/sec (median output speed, first-party API); TTFT 102.1s; list price $10/50 per 1M in/out
Artificial Analysis output speed	68 tok/s	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	tokens/sec (median output speed, first-party API); TTFT 262.3s; list price $10/50 per 1M in/out
ArXivMath (no tools)	82.9%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	Aug 2026
ArXivMath (with tools)	92.1%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	Aug 2026
AutomationBench	31.4%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	AutomationBench (Zapier; end-to-end business workflows across connected apps)
Chartography (with tools)	88.4%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—
CursorBench	45.1%	LowFable 5.1 Low	Independent testIndependentCursor (CursorBench) ↗	1 Oct 2026	Cursor agent	CursorBench 4.0 (tasks from real Cursor sessions; v4.0 introduced 2026-09-10); rank 15; $5.44/task; 34,795 tokens/task; 51 steps/task
CursorBench	45.1%	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Cursor agent	CursorBench 4.0, run end-to-end by Cursor in its production agent harness; vendor-reported cost/task $5.44
CursorBench	46.8%	MediumFable 5.1 Medium	Independent testIndependentCursor (CursorBench) ↗	1 Oct 2026	Cursor agent	CursorBench 4.0 (tasks from real Cursor sessions; v4.0 introduced 2026-09-10); rank 11; $7.05/task; 45,411 tokens/task; 63 steps/task
CursorBench	46.8%	Mediumadaptive thinking, effort=med	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Cursor agent	CursorBench 4.0, run end-to-end by Cursor in its production agent harness; vendor-reported cost/task $7.05
CursorBench	49.2%	HighFable 5.1 High	Independent testIndependentCursor (CursorBench) ↗	1 Oct 2026	Cursor agent	CursorBench 4.0 (tasks from real Cursor sessions; v4.0 introduced 2026-09-10); rank 9; $9.08/task; 58,438 tokens/task; 77 steps/task
CursorBench	49.2%	Highadaptive thinking, effort=high	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Cursor agent	CursorBench 4.0, run end-to-end by Cursor in its production agent harness; vendor-reported cost/task $9.08
CursorBench	51.6%	Extra highFable 5.1 Extra High	Independent testIndependentCursor (CursorBench) ↗	1 Oct 2026	Cursor agent	CursorBench 4.0 (tasks from real Cursor sessions; v4.0 introduced 2026-09-10); rank 8; $13.01/task; 87,294 tokens/task; 101 steps/task
CursorBench	51.6%	Extra highadaptive thinking, effort=xhigh	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Cursor agent	CursorBench 4.0, run end-to-end by Cursor in its production agent harness; vendor-reported cost/task $13.01
CursorBench	51.8%	MaxFable 5.1 Max	Independent testIndependentCursor (CursorBench) ↗	1 Oct 2026	Cursor agent	CursorBench 4.0 (tasks from real Cursor sessions; v4.0 introduced 2026-09-10); rank 7; $17.28/task; 117,236 tokens/task; 128 steps/task
CursorBench	51.8%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Cursor agent	CursorBench 4.0, run end-to-end by Cursor in its production agent harness; vendor-reported cost/task $17.28
CursorBench 3.2.0	66.2%	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Cursor agent	CursorBench 3.2.0 (Cursor production harness); not comparable to CursorBench 4.0; vendor-reported cost/task $2.9
CursorBench 3.2.0	68.0%	Mediumadaptive thinking, effort=med	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Cursor agent	CursorBench 3.2.0 (Cursor production harness); not comparable to CursorBench 4.0; vendor-reported cost/task $3.53
CursorBench 3.2.0	69.4%	Highadaptive thinking, effort=high	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Cursor agent	CursorBench 3.2.0 (Cursor production harness); not comparable to CursorBench 4.0; vendor-reported cost/task $4.8
CursorBench 3.2.0	72.8%	Extra highadaptive thinking, effort=xhigh	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Cursor agent	CursorBench 3.2.0 (Cursor production harness); not comparable to CursorBench 4.0; vendor-reported cost/task $6.96
CursorBench 3.2.0	73.4%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Cursor agent	CursorBench 3.2.0 (Cursor production harness); not comparable to CursorBench 4.0; vendor-reported cost/task $9.64
DeepSWE	63.1%	Extra highClaude Fable 5.1 XHigh + SWE-2 Medium	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	Devin Fusion CLI	AA Coding Agent Index component; pass@1 avg of 3 attempts
DeepSWE	64.3%	MaxFable 5.1 (max) (with fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	Claude Code	AA Coding Agent Index component; pass@1 avg of 3 attempts
DeepSWE	67.4%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—
Design Arena (all categories)	1342	Defaultclaude-fable-5-1	Independent testIndependentDesign Arena ↗	1 Oct 2026	—	Bradley-Terry Elo; ±3.8 SE; 9335 battles; win rate 59.3%
Design Arena (fullstack)	1324	Defaultclaude-fable-5-1	Independent testIndependentDesign Arena ↗	1 Oct 2026	—	Bradley-Terry Elo; ±15.9 SE; 509 battles; win rate 57.2%
Epoch Capabilities Index	164.8	Default	Independent testIndependentEpoch AI ↗	1 Oct 2026	—	Epoch Capabilities Index; 90% CI 161.8-169.1; best of listed model versions
EQ-Bench Creative Writing v3 (Elo)	2162	Default*claude-fable-5-1	Independent testIndependentEQ-Bench ↗	1 Oct 2026	—	leaderboard rank 2; rubric score 16.95/20; slop 8.16; avg length 5841 chars; Elo judged by Claude Sonnet 4.6; effort not stated by EQ-Bench; marked new (*)
FrontierCode	49.8%	Loweffort=low	Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗	22 Sep 2026	—	FrontierCode 1.1 Main (Cognition); GPT models run in Codex CLI; vendor-estimated API cost/task $2.38; competitor number as plotted by OpenAI
FrontierCode	52.8%	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code	FrontierCode v1.1 Main (150 tasks, run & scored by Cognition; composite of correctness + mergeability rubric; penalizes out-of-scope edits); Fable 5.1 adds unrequested out-of-scope edits at higher effort; vendor-reported cost/task $2.47
FrontierCode	50.9%	Mediumclaude-fable-5-1_medium	Independent testIndependentFrontierCode (Cognition) via Epoch AI ↗	1 Oct 2026	Claude Code	read from Epoch AI benchmark_data.zip (frontiercode_external.csv); original leaderboard https://cognition.com/frontiercode; harness claude-code; Mean@5
FrontierCode	50.9%	Mediumadaptive thinking, effort=med	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code	FrontierCode v1.1 Main (150 tasks, run & scored by Cognition; composite of correctness + mergeability rubric; penalizes out-of-scope edits); Fable 5.1 adds unrequested out-of-scope edits at higher effort; vendor-reported cost/task $3.28
FrontierCode	50.3%	Highadaptive thinking, effort=high	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code	FrontierCode v1.1 Main (150 tasks, run & scored by Cognition; composite of correctness + mergeability rubric; penalizes out-of-scope edits); Fable 5.1 adds unrequested out-of-scope edits at higher effort; vendor-reported cost/task $5.27
FrontierCode	48.7%	Extra highadaptive thinking, effort=xhigh	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code	FrontierCode v1.1 Main (150 tasks, run & scored by Cognition; composite of correctness + mergeability rubric; penalizes out-of-scope edits); Fable 5.1 adds unrequested out-of-scope edits at higher effort; vendor-reported cost/task $9.27
FrontierCode	50.3%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code	FrontierCode v1.1 Main (150 tasks, run & scored by Cognition; composite of correctness + mergeability rubric; penalizes out-of-scope edits); Fable 5.1 adds unrequested out-of-scope edits at higher effort; vendor-reported cost/task $12.82
FrontierCode v1.1 (Extended)	63.6%	Mediumadaptive thinking, effort=medium (best)	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Claude Code	Fable 5.1 peaks at medium on FrontierCode; adds unrequested out-of-scope edits at higher effort
FrontierMath (Tiers 1-3)	90.2%	Maxclaude-fable-5-1_max	Independent testIndependentEpoch AI ↗	1 Sep 2026	—	FrontierMath Tiers 1-3 (v2), Epoch-run; stderr 1.8pt; data https://epoch.ai/data/benchmark_data.zip (frontiermath_tiers_1_3_v2.csv)
FrontierMath Tier 4	87.8%	Maxclaude-fable-5-1_max	Independent testIndependentEpoch AI ↗	1 Sep 2026	—	FrontierMath Tier 4 (v2), Epoch-run; stderr 5.2pt; data https://epoch.ai/data/benchmark_data.zip (frontiermath_tier_4_v2.csv)
FrontierSWE v2	56.3%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Proximal harness	reported as 0.57 on a 0-1 scale (Opus 5.5 card later lists 56.3%)
GDPval-AA v2	1853	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	GDPval-AA v2 Elo
GDPval-AA v2.1	1450	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	GDPval-AA v2.1 (Artificial Analysis, Elo from blind pairwise comparisons, 44 occupations); vendor-reported cost/task $1.41
GDPval-AA v2.1	1536	Mediumadaptive thinking, effort=med	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	GDPval-AA v2.1 (Artificial Analysis, Elo from blind pairwise comparisons, 44 occupations); vendor-reported cost/task $2.17
GDPval-AA v2.1	1617	Highadaptive thinking, effort=high	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	GDPval-AA v2.1 (Artificial Analysis, Elo from blind pairwise comparisons, 44 occupations); vendor-reported cost/task $3.43
GDPval-AA v2.1	1721	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	95% CI 1695.77-1746.08
GDPval-AA v2.1	1721	Extra highadaptive thinking, effort=xhigh	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	GDPval-AA v2.1 (Artificial Analysis, Elo from blind pairwise comparisons, 44 occupations); vendor-reported cost/task $7.09
GDPval-AA v2.1	1735	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	95% CI 1710.09-1759.25
GDPval-AA v2.1	1735	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	GDPval-AA v2.1 (Artificial Analysis, Elo from blind pairwise comparisons, 44 occupations); vendor-reported cost/task $9.59
GPQA Diamond	88.1%	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-low)
GPQA Diamond	88.6%	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-medium)
GPQA Diamond	90.6%	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-high)
GPQA Diamond	93.4%	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-xhigh)
GPQA Diamond	93.7%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1)
GPQA Diamond	93.4%	Maxeffort=max	Independent testIndependentVals.ai ↗	1 Sep 2026	—	±1.863 stderr; $0.205/test
GSO	88.2%	Extra highreasoning_effort=xhigh	Independent testIndependentGSO ↗	27 Sep 2026	OpenHands	Opt@1; hack-controlled score 87.25; data https://gso-bench.github.io/assets/leaderboard.json
Harvey LAB-AA	93.0%	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	criteria pass rate, AA-run
Harvey LAB-AA	93.3%	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	criteria pass rate, AA-run
Harvey LAB-AA	93.0%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	criteria pass rate, AA-run
HealthBench Professional	62.1%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—
HLE Diamond	51.3%	Defaultclaude-fable-5-1	Independent testIndependentScale AI SEAL ↗	1 Oct 2026	—	rank 3 (Scale rank accounts for CI); ±3.1; entry added 2026-04-10
Humanity's Last Exam	48.9%	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-low)
Humanity's Last Exam	53.2%	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam without tools; vendor-reported cost/task $0.3
Humanity's Last Exam	53.8%	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-medium)
Humanity's Last Exam	55.9%	Mediumadaptive thinking, effort=med	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam without tools; vendor-reported cost/task $0.46
Humanity's Last Exam	55.9%	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-high)
Humanity's Last Exam	58.0%	Highadaptive thinking, effort=high	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam without tools; vendor-reported cost/task $0.75
Humanity's Last Exam	58.7%	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-xhigh)
Humanity's Last Exam	46.5%	Extra highFable 5.1 (xhigh)	Independent testIndependentScale AI SEAL ↗	1 Oct 2026	—	rank 2 (Scale rank accounts for CI); ±2; entry added 2026-09-03
Humanity's Last Exam	60.4%	Extra highadaptive thinking, effort=xhigh	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam without tools; vendor-reported cost/task $1.53
Humanity's Last Exam	59.1%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1)
Humanity's Last Exam	60.9%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam without tools; vendor-reported cost/task $2.23
Humanity's Last Exam (with tools)	60.0%	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam with tools (web search/fetch, code execution); vendor-reported cost/task $0.52
Humanity's Last Exam (with tools)	63.0%	Mediumadaptive thinking, effort=med	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam with tools (web search/fetch, code execution); vendor-reported cost/task $0.67
Humanity's Last Exam (with tools)	64.8%	Highadaptive thinking, effort=high	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam with tools (web search/fetch, code execution); vendor-reported cost/task $1.05
Humanity's Last Exam (with tools)	65.1%	Extra highadaptive thinking, effort=xhigh	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam with tools (web search/fetch, code execution); vendor-reported cost/task $2.28
Humanity's Last Exam (with tools)	65.0%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	Humanity's Last Exam with tools (web search/fetch, code execution); vendor-reported cost/task $3.2
IOI (Vals)	90.8%	Maxeffort=max	Independent testIndependentVals.ai ↗	29 Sep 2026	—	±4.647 stderr; $11.197/test
LegalBench (Vals)	88.5%	Maxeffort=max	Independent testIndependentVals.ai ↗	29 Sep 2026	—	vals id anthropic/claude-fable-5-1; rank 2/149; ±0.42 stderr; $0.051606/test
LiveCodeBench	90.5%	Maxeffort=max	Independent testIndependentVals.ai ↗	1 Sep 2026	—	±0.863 stderr; $0.261/test
LLM Creative Story-Writing Benchmark (Lech Mazur)	3.8	HighClaude Fable 5.1 (high)	Independent testIndependentLech Mazur (LLM Creative Story-Writing Benchmark) ↗	28 Sep 2026	—	rank 1/56; Thurstone comparison score (centered at 0); est. win chance 93%; 95% bootstrap 3.702 to 3.896
LMArena Code Arena (WebDev)	1751	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	Code Arena \| WebDev overall (agentic web-dev, raw); rank 4 (CI rank 3-4); 95% CI 1741-1760; 6137 votes
LMArena Text - Coding category	1533	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	text arena coding category, style control on; rank 17 (CI rank 2-48); 95% CI 1520-1545; 2254 votes
LMArena Text - Creative Writing	1482	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 10 (rank range 4-26); 95% CI 1468.8-1494.5; 2506 votes; style-controlled
LMArena Text - Expert	1519	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 16 (rank range 3-62); 95% CI 1499.9-1538.6; 965 votes; style-controlled
LMArena Text - Hard Prompts	1522	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 7 (rank range 2-19); 95% CI 1513.6-1529.4; 6630 votes; style-controlled
LMArena Text - Instruction Following	1498	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 7 (rank range 3-20); 95% CI 1486.9-1508.5; 3380 votes; style-controlled
LMArena Text - Longer Query	1511	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 7 (rank range 2-21); 95% CI 1502.0-1520.8; 4705 votes; style-controlled
LMArena Text - Multi-Turn	1487	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 25 (rank range 7-74); 95% CI 1471.1-1502.7; 1474 votes; style-controlled
LMArena Text - Non-English	1495	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 3 (rank range 1-13); 95% CI 1487.6-1503.2; 6689 votes; style-controlled
LMArena Text - Occupational: Business, Management & Finance	1487	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 15 (rank range 3-48); 95% CI 1473.7-1499.8; 2102 votes; style-controlled
LMArena Text - Occupational: Legal & Government	1494	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 19 (rank range 1-74); 95% CI 1474.2-1514.4; 918 votes; style-controlled
LMArena Text - Occupational: Writing, Literature & Language	1493	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	rank 7 (rank range 2-19); 95% CI 1481.5-1504.0; 3159 votes; style-controlled
LMArena Text (overall)	1501	Maxclaude-fable-5.1-max	Independent testIndependentLMArena ↗	30 Sep 2026	—	text overall, style control on; rank 6 (CI rank 2-14); 95% CI 1495-1508; 11241 votes
MCP Atlas	87.2%	DefaultFable 5.1	Independent testIndependentScale AI SEAL ↗	1 Oct 2026	—	rank 1 (Scale rank accounts for CI); ±2.05; entry added 2026-09-04
OfficeQA Pro	69.0%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—
OSWorld 2.0	77.9%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	partial score; August 2026 task release; safeguard interventions scored 0
OSWorld 2.0 (strict pass rate)	41.7%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	—	strict pass rate; August 2026 task release
OSWorld 2.1 (partial score)	80.7%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	partial score; re-evaluated by Anthropic under Opus 5.5 conditions
OSWorld 2.1 (strict pass rate)	42.8%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—
PRBench Finance (Scale)	50.8%	DefaultFable 5.1	Independent testIndependentScale AI (SEAL) ↗	3 Sep 2026	—	Scale rank 7; ±0.22 CI
PRBench Legal (Scale)	51.6%	DefaultFable 5.1	Independent testIndependentScale AI (SEAL) ↗	3 Sep 2026	—	Scale rank 5; ±0.24 CI
ProgramBench (fully resolved)	7.0%	Maxeffort=max	Independent testIndependentVals.ai ↗	29 Sep 2026	mini-SWE-agent	±1.809 stderr; $58.517/test; strict fully-resolved rate
ProgramBench (fully resolved)	87.6%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	mini-swe-agent
SciCode	56.7%	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-low)
SciCode	56.4%	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-medium)
SciCode	58.7%	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-high)
SciCode	60.9%	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-xhigh)
SciCode	63.1%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1)
SimpleBench	86.6%	DefaultClaude Fable 5.1	Independent testIndependentSimpleBench ↗	3 Sep 2026	—	AVG@5, temp 0.7; rank 2nd; human baseline 83.7%; data https://simple-bench.com/static/js/leaderboard-data.js
SimpleQA Verified	70.8%	Maxclaude-fable-5-1_max	Independent testIndependentEpoch AI ↗	1 Sep 2026	—	Epoch-run (no tools); ±1.44 stderr
SkillsBench	61.5%	Maxeffort=max	Independent testIndependentVals.ai ↗	27 Sep 2026	OpenHands	±4.76 stderr; $4.517/test
SWE Atlas - Codebase QnA	65.9%	Extra highClaude Fable 5.1 XHigh + SWE-2 Medium	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	Devin Fusion CLI	AA Coding Agent Index component; pass@1 avg of 3 attempts
SWE Atlas - Codebase QnA	60.0%	Extra highFable-5.1 (Claude Code) xHigh*	Independent testIndependentScale AI SEAL ↗	1 Oct 2026	Claude Code	rank 1 (Scale rank accounts for CI); ±4.85; entry added 2026-03-02; * = Scale footnote (see page)
SWE Atlas - Codebase QnA	64.8%	MaxFable 5.1 (max) (with fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	Claude Code	AA Coding Agent Index component; pass@1 avg of 3 attempts
SWE Atlas - Refactoring	56.7%	Extra highFable-5.1 (Claude Code) xHigh	Independent testIndependentScale AI SEAL ↗	1 Oct 2026	Claude Code	rank 1 (Scale rank accounts for CI); ±6.52; entry added 2026-09-02
SWE Atlas - Test Writing	67.0%	Extra highFable-5.1 (Claude Code) xHigh*	Independent testIndependentScale AI SEAL ↗	1 Oct 2026	Claude Code	rank 1 (Scale rank accounts for CI); ±5.33; entry added 2026-09-02; * = Scale footnote (see page)
SWE-bench Multilingual	89.1%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—
SWE-bench Multimodal	54.7%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—
SWE-bench Pro (Anthropic internal subset)	88.6%	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	1 Oct 2026	—	Anthropic internal 478-problem SWE-bench Pro subset (not comparable to public leaderboard), priced as billed; $0.54 per solved task
SWE-bench Pro (Anthropic internal subset)	92.3%	Highadaptive thinking, effort=high (default)	Maker's own figureVendor-reportedAnthropic ↗	1 Oct 2026	—	Anthropic internal 478-problem SWE-bench Pro subset (not comparable to public leaderboard), priced as billed; $1.19 per solved task
SWE-Bench Pro (public, v1)	81.2%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—
SWE-Bench Pro V2 (full)	99.1%	HighFable 5.1 (Claude Code) high	Independent testIndependentScale AI SEAL ↗	1 Oct 2026	Claude Code	rank 2 (Scale rank accounts for CI); ±0.5; entry added 2026-09-22; SWE-Bench Pro V2 (642 tasks, locked protocol, released 2026-09-22)
SWE-Bench Pro V2 (hard)	92.2%	HighFable 5.1 (Claude Code) high	Independent testIndependentScale AI SEAL ↗	1 Oct 2026	Claude Code	rank 2 (Scale rank accounts for CI); ±0; entry added 2026-09-22; SWE-Bench Pro V2 (642 tasks, locked protocol, released 2026-09-22)
Terminal-Bench 2.1	85.0%	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-low)
Terminal-Bench 2.1	88.0%	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-medium)
Terminal-Bench 2.1	89.9%	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-high)
Terminal-Bench 2.1	85.0%	Higheffort=high	Independent testIndependentVals.ai ↗	27 Sep 2026	Terminus 2	±0.991 stderr; $3.075/test
Terminal-Bench 2.1	91.0%	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-xhigh)
Terminal-Bench 2.1	91.4%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1)
Terminal-Bench 4.0	43.3%	Low	Independent testIndependentTerminal-Bench ↗	1 Oct 2026	Claude Code	TB 4.0.0 (66 tasks); ±3.61 95% CI; 330 trials; total run cost $2359; model release 2026-09-01
Terminal-Bench 4.0	40.4%	LowClaude Fable 5.1 (Adaptive Reasoning, Low Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-low)
Terminal-Bench 4.0	40.2%	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code (--bare)	Terminal-Bench 4.0 (66 tasks); Anthropic internal runs, 5-15 trials/task; production safeguards on; vendor-reported cost/task $5.7
Terminal-Bench 4.0	53.9%	Medium	Independent testIndependentTerminal-Bench ↗	1 Oct 2026	Claude Code	TB 4.0.0 (66 tasks); ±3.39 95% CI; 330 trials; total run cost $2833; model release 2026-09-01
Terminal-Bench 4.0	44.9%	MediumClaude Fable 5.1 (Adaptive Reasoning, Medium Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-medium)
Terminal-Bench 4.0	43.4%	Mediumadaptive thinking, effort=med	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code (--bare)	Terminal-Bench 4.0 (66 tasks); Anthropic internal runs, 5-15 trials/task; production safeguards on; vendor-reported cost/task $7.8
Terminal-Bench 4.0	54.5%	High	Independent testIndependentTerminal-Bench ↗	1 Oct 2026	Claude Code	TB 4.0.0 (66 tasks); ±3.44 95% CI; 330 trials; total run cost $3985; model release 2026-09-01
Terminal-Bench 4.0	52.0%	HighClaude Fable 5.1 (Adaptive Reasoning, High Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-high)
Terminal-Bench 4.0	49.4%	Highadaptive thinking, effort=high	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code (--bare)	Terminal-Bench 4.0 (66 tasks); Anthropic internal runs, 5-15 trials/task; production safeguards on; vendor-reported cost/task $10.5
Terminal-Bench 4.0	57.9%	Extra high	Independent testIndependentTerminal-Bench ↗	1 Oct 2026	Claude Code	TB 4.0.0 (66 tasks); ±3.36 95% CI; 330 trials; total run cost $4872; model release 2026-09-01
Terminal-Bench 4.0	56.1%	Extra highClaude Fable 5.1 XHigh + SWE-2 Medium	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	Devin Fusion CLI	AA Coding Agent Index component run in the vendor agent harness; pass@1 avg of 3 attempts
Terminal-Bench 4.0	55.1%	Extra highClaude Fable 5.1 (Adaptive Reasoning, Xhigh Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1-xhigh)
Terminal-Bench 4.0	51.3%	Extra highadaptive thinking, effort=xhigh	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code (--bare)	Terminal-Bench 4.0 (66 tasks); Anthropic internal runs, 5-15 trials/task; production safeguards on; vendor-reported cost/task $15.8
Terminal-Bench 4.0	58.1%	Maxeffort=max	Independent testIndependentVals.ai ↗	29 Sep 2026	mini-SWE-agent	±3.312 stderr; $17.175/test
Terminal-Bench 4.0	57.9%	Max	Independent testIndependentTerminal-Bench ↗	1 Oct 2026	Claude Code	TB 4.0.0 (66 tasks); ±3.76 95% CI; 330 trials; total run cost $6244; model release 2026-09-01
Terminal-Bench 4.0	57.6%	MaxFable 5.1 (max) (with fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	Claude Code	AA Coding Agent Index component run in the vendor agent harness; pass@1 avg of 3 attempts
Terminal-Bench 4.0	52.0%	MaxClaude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)	Independent testIndependentArtificial Analysis ↗	1 Oct 2026	—	AA-run evaluation (AA slug claude-fable-5-1)
Terminal-Bench 4.0	55.8%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	Claude Code (--bare)	Terminal-Bench 4.0 (66 tasks); Anthropic internal runs, 5-15 trials/task; production safeguards on; vendor-reported cost/task $19.5
Terminal-Bench-Science 0.1	26.3%	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Claude Code (--bare)	Terminal-Bench-Science 0.1 (70 tasks); SE ±3.5-4.5; vendor-reported cost/task $11.1
Terminal-Bench-Science 0.1	35.7%	Mediumadaptive thinking, effort=med	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Claude Code (--bare)	Terminal-Bench-Science 0.1 (70 tasks); SE ±3.5-4.5; vendor-reported cost/task $14.9
Terminal-Bench-Science 0.1	40.0%	Highadaptive thinking, effort=high	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Claude Code (--bare)	Terminal-Bench-Science 0.1 (70 tasks); SE ±3.5-4.5; vendor-reported cost/task $20.3
Terminal-Bench-Science 0.1	49.5%	Extra highadaptive thinking, effort=xhigh	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Claude Code (--bare)	Terminal-Bench-Science 0.1 (70 tasks); SE ±3.5-4.5; vendor-reported cost/task $31.8
Terminal-Bench-Science 0.1	52.6%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	1 Sep 2026	Claude Code (--bare)	Terminal-Bench-Science 0.1 (70 tasks); SE ±3.5-4.5; vendor-reported cost/task $37.9
Toolathlon	77.8%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	internal harness, Pass@1
Vals Code Migration	54.6%	Maxeffort=max	Independent testIndependentVals.ai ↗	29 Sep 2026	—	±4.812 stderr; $70.973/test
Vals Index	65.8%	Maxeffort=max	Independent testIndependentVals.ai ↗	30 Sep 2026	—	±1.115 stderr; $28.712/test
Vals Legal Research Bench	55.3%	Maxeffort=max	Independent testIndependentVals.ai ↗	29 Sep 2026	—	vals id anthropic/claude-fable-5-1; rank 3/72; ±3.456 stderr; $23.055789/test
Vals Public Benefits Bench v1.1	74.9%	Maxeffort=max	Independent testIndependentVals.ai ↗	29 Sep 2026	—	vals id anthropic/claude-fable-5-1; rank 2/45; ±1.128 stderr; $7.481573/test
Vals SRE Bench	22.9%	Maxeffort=max	Independent testIndependentVals.ai ↗	29 Sep 2026	—	±2.601 stderr; $32.232/test
Vals TaxEval v2	76.0%	Maxeffort=max	Independent testIndependentVals.ai ↗	1 Sep 2026	—	vals id anthropic/claude-fable-5-1; rank 9/145; ±0.83 stderr; $0.346549/test
Vals Vibe Code Bench	90.3%	Maxeffort=max	Independent testIndependentVals.ai ↗	29 Sep 2026	OpenHands	±1.566 stderr; $33.368/test
WANDR	63.3%	Lowadaptive thinking, effort=low	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	WANDR (Perplexity wide-search benchmark), soft F1; Anthropic-modified setup (offline search/fetch, 980k token budget), not comparable to Perplexity numbers; vendor-reported cost/task $23.25
WANDR	64.5%	Mediumadaptive thinking, effort=med	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	WANDR (Perplexity wide-search benchmark), soft F1; Anthropic-modified setup (offline search/fetch, 980k token budget), not comparable to Perplexity numbers; vendor-reported cost/task $27.41
WANDR	66.7%	Highadaptive thinking, effort=high	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	WANDR (Perplexity wide-search benchmark), soft F1; Anthropic-modified setup (offline search/fetch, 980k token budget), not comparable to Perplexity numbers; vendor-reported cost/task $33.62
WANDR	67.7%	Extra highadaptive thinking, effort=xhigh	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	WANDR (Perplexity wide-search benchmark), soft F1; Anthropic-modified setup (offline search/fetch, 980k token budget), not comparable to Perplexity numbers; vendor-reported cost/task $42.81
WANDR	68.7%	Maxadaptive thinking, effort=max	Maker's own figureVendor-reportedAnthropic ↗	22 Sep 2026	—	WANDR (Perplexity wide-search benchmark), soft F1; Anthropic-modified setup (offline search/fetch, 980k token budget), not comparable to Perplexity numbers; vendor-reported cost/task $49.02