Models · Anthropic · Out sinceReleased 9 Jun 2026
Claude Fable 5#3 for writing.#3 for writing, best at high effort.
Available on OpenRouterOn OpenRouterAvailable nowGenerally availableClosed (can't be downloaded)Proprietary
Claude Fable 5 is made by Anthropic. Among the models we track it ranks #3 for writing, #7 for research and analysis, #8 for coding. It's expensive to use.
API id claude-fable-5. Same model as Claude Mythos 5 with safeguards (fallback to Opus 4.8). Default effort high; no per-message effort.
- 78.2 / 100 · #3
- 79.8 / 100 · #7
- 71.8 / 100 · #8
- Expensive$10 / $50
- $20
- 1M
- 128K
- 124 (79 independent79 indep.)
- 9 Jun 2026
- Anthropic
- 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.
LowMediumHighExtra highMax
Highlighted: where it did best for writing (High thinking).
Highlighted: dominant setting in its writing composite.
OpenRouter is a service that gives access to many AI models in one place. This model is listed there as anthropic/claude-fable-5.
Route it as anthropic/claude-fable-5 at $10 in / $50 out per 1M tokens, 1M context. Listed since 9 Jun 2026.
include_reasoningmax_completion_tokensmax_tokensreasoningreasoning_effortresponse_formatstopstructured_outputstool_choicetoolsverbosity
Thinking level
Reasoning effort
How long should Claude Fable 5Where Claude Fable 5think?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
DeepSWE
Best at Extra high, worse abovePeaks at Extra high
CursorBench 3.2.0
Best at MaxBest at Max
Terminal-Bench 2.1
Best at MaxBest at Max
Terminal-Bench-Science 0.1
Best at High, worse abovePeaks at High
ARC-AGI-2
Best at MaxBest at Max
Humanity's Last Exam
Best at Extra high, worse abovePeaks at Extra high
Humanity's Last Exam (with tools)
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.
| NotesNotes | ||||||
|---|---|---|---|---|---|---|
| AA Analyst Agent | 48.8% | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA-run; scores move in 1.25-pt steps (small task set) |
| AA-Briefcase | 1574 | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 24 Jul 2026 | — | |
| Agents' Last Exam | 48.7% | Extra higheffort=xhigh | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 22 Sep 2026 | — | Agents' Last Exam V1 (long-horizon professional workflows, 55 sub-industries); vendor-estimated API cost/task $28.5542; competitor number as plotted by OpenAI ("Claude Fable 5 w/ Opus 4.8 fallback") |
| Agents' Last Exam | 41.3% | Defaultadaptive (as labeled) | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 22 Sep 2026 | — | Agents' Last Exam V1 (long-horizon professional workflows, 55 sub-industries); vendor-estimated API cost/task $15.2285; competitor number as plotted by OpenAI ("Claude Fable 5 w/ Opus 4.8 fallback") |
| ARC-AGI-1 | 98.5% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | |
| ARC-AGI-2 | 87.5% | HighClaude Fable 5 (High) | Independent testIndependentARC Prize ↗ | 30 Sep 2026 | — | semi-private set; $2.44/task; data https://arcprize.org/media/data/leaderboard/v2.json |
| ARC-AGI-2 | 88.3% | Extra highClaude Fable 5 (XHigh) | Independent testIndependentARC Prize ↗ | 30 Sep 2026 | — | semi-private set; $3.00/task; data https://arcprize.org/media/data/leaderboard/v2.json |
| ARC-AGI-2 | 89.2% | MaxClaude Fable 5 (Max) | Independent testIndependentARC Prize ↗ | 30 Sep 2026 | — | semi-private set; $5.45/task; data https://arcprize.org/media/data/leaderboard/v2.json |
| ARC-AGI-2 | 89.2% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | |
| Artificial Analysis Coding Agent Index | 58.6 | Loweffort=low | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 3 Sep 2026 | — | Artificial Analysis Coding Agent Index v1.4 (index score 0-100); vendor-estimated API cost/task $3.17; competitor number as plotted by OpenAI |
| Artificial Analysis Coding Agent Index | 63.2 | Mediumeffort=medium | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 3 Sep 2026 | — | Artificial Analysis Coding Agent Index v1.4 (index score 0-100); vendor-estimated API cost/task $4.74; competitor number as plotted by OpenAI |
| Artificial Analysis Coding Agent Index | 65.1 | Higheffort=high | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 3 Sep 2026 | — | Artificial Analysis Coding Agent Index v1.4 (index score 0-100); vendor-estimated API cost/task $5.97; competitor number as plotted by OpenAI |
| Artificial Analysis Coding Agent Index | 66 | Extra higheffort=xhigh | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 3 Sep 2026 | — | Artificial Analysis Coding Agent Index v1.4 (index score 0-100); vendor-estimated API cost/task $8.53; competitor number as plotted by OpenAI |
| Artificial Analysis Coding Agent Index | 67.2 | Maxeffort=max | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 3 Sep 2026 | — | Artificial Analysis Coding Agent Index v1.4 (index score 0-100); vendor-estimated API cost/task $11.69; competitor number as plotted by OpenAI |
| Artificial Analysis Intelligence Index | 49.6 | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA slug claude-fable-5; list price $10/50 per 1M in/out; cost to run AA Intelligence Index $8.75/task (AA marks this variant deprecated) |
| Artificial Analysis output speed | 63 tok/s | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | tokens/sec (median output speed, first-party API); TTFT 96.0s; list price $10/50 per 1M in/out (AA marks this variant deprecated) |
| AutomationBench | 17.1% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | read from launch-post benchmark table image (values printed in table) |
| Blueprint-Bench 2 | 38.6% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 9 Jun 2026 | — | read from launch-post benchmark table image (values printed in table) |
| BrowseComp | 87.4% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 24 Jul 2026 | — | read from launch-post benchmark table image (values printed in table) |
| CursorBench (pre-3.2, legacy) | 72.9% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 9 Jun 2026 | Cursor agent | CursorBench version as of June 2026; leads at every effort from medium up |
| CursorBench 3.2.0 | 62.1% | 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 $4.46 |
| CursorBench 3.2.0 | 65.2% | 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 $6.8 |
| CursorBench 3.2.0 | 66.5% | 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 $8.77 |
| CursorBench 3.2.0 | 68.4% | 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 $11.73 |
| CursorBench 3.2.0 | 70.5% | 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 $17.32 |
| DeepSWE | 59.6% | Lowclaude-fable-5_low | Independent testIndependentDeepSWE (Datacurve) via Epoch AI ↗ | 1 Oct 2026 | mini-SWE-agent | read from Epoch AI benchmark_data.zip (deepswe_external.csv); original leaderboard https://deepswe.datacurve.ai/; harness mini-swe-agent; pass@4 81.4%; ±2.8; 4 runs; $3.76/task |
| DeepSWE | 59.6% | Loweffort=low | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 22 Sep 2026 | — | DeepSWE v1.1 (113 long-horizon SWE tasks); GPT models run in Codex; vendor-estimated API cost/task $3.7579; competitor number as plotted by OpenAI |
| DeepSWE | 65.4% | Mediumclaude-fable-5_medium | Independent testIndependentDeepSWE (Datacurve) via Epoch AI ↗ | 1 Oct 2026 | mini-SWE-agent | read from Epoch AI benchmark_data.zip (deepswe_external.csv); original leaderboard https://deepswe.datacurve.ai/; harness mini-swe-agent; pass@4 83.2%; ±4.4; 4 runs; $6.09/task |
| DeepSWE | 65.4% | Mediumeffort=medium | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 22 Sep 2026 | — | DeepSWE v1.1 (113 long-horizon SWE tasks); GPT models run in Codex; vendor-estimated API cost/task $6.0882; competitor number as plotted by OpenAI |
| DeepSWE | 68.6% | Highclaude-fable-5_high | Independent testIndependentDeepSWE (Datacurve) via Epoch AI ↗ | 1 Oct 2026 | mini-SWE-agent | read from Epoch AI benchmark_data.zip (deepswe_external.csv); original leaderboard https://deepswe.datacurve.ai/; harness mini-swe-agent; pass@4 86.7%; ±1.1; 4 runs; $9.18/task |
| DeepSWE | 68.6% | Higheffort=high | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 22 Sep 2026 | — | DeepSWE v1.1 (113 long-horizon SWE tasks); GPT models run in Codex; vendor-estimated API cost/task $9.1776; competitor number as plotted by OpenAI |
| DeepSWE | 69.9% | Extra higheffort=xhigh | Independent testIndependentOpenAI (competitor result in OpenAI launch post) ↗ | 22 Sep 2026 | — | DeepSWE v1.1 (113 long-horizon SWE tasks); GPT models run in Codex; vendor-estimated API cost/task $13.4145; competitor number as plotted by OpenAI |
| DeepSWE | 69.9% | Extra highclaude-fable-5_xhigh | Independent testIndependentDeepSWE (Datacurve) via Epoch AI ↗ | 1 Oct 2026 | mini-SWE-agent | read from Epoch AI benchmark_data.zip (deepswe_external.csv); original leaderboard https://deepswe.datacurve.ai/; harness mini-swe-agent; pass@4 88.5%; ±3.2; 4 runs; $13.41/task |
| DeepSWE | 69.7% | Maxclaude-fable-5_max | Independent testIndependentDeepSWE (Datacurve) via Epoch AI ↗ | 1 Oct 2026 | mini-SWE-agent | read from Epoch AI benchmark_data.zip (deepswe_external.csv); original leaderboard https://deepswe.datacurve.ai/; harness mini-swe-agent; pass@4 84.1%; ±4.0; 4 runs; $21.63/task |
| DeepSWE | 69.7% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 24 Jul 2026 | — | read from launch-post benchmark table image (values printed in table) |
| Design Arena (all categories) | 1317 | Defaultclaude-fable-5 | Independent testIndependentDesign Arena ↗ | 1 Oct 2026 | — | Bradley-Terry Elo; ±2.7 SE; 20185 battles; win rate 57.7% |
| Design Arena (fullstack) | 1278 | Defaultclaude-fable-5 | Independent testIndependentDesign Arena ↗ | 1 Oct 2026 | — | Bradley-Terry Elo; ±8.2 SE; 2118 battles; win rate 58.2% |
| Epoch Capabilities Index | 162.2 | Default | Independent testIndependentEpoch AI ↗ | 1 Oct 2026 | — | Epoch Capabilities Index; 90% CI 159.6-166.1; best of listed model versions |
| EQ-Bench Creative Writing v3 (Elo) | 1943 | Defaultclaude-fable-5 | Independent testIndependentEQ-Bench ↗ | 1 Oct 2026 | — | leaderboard rank 11; rubric score 16.81/20; slop 10.28; avg length 5887 chars; Elo judged by Claude Sonnet 4.6; effort not stated by EQ-Bench |
| Frontier-Bench v0.1 | 33.7% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 24 Jul 2026 | — | read from launch-post benchmark table image (values printed in table) |
| FrontierCode | 53.5% | Extra highadaptive thinking, effort=xhigh (best) | Maker's own figureVendor-reportedAnthropic ↗ | 24 Jul 2026 | Claude Code | FrontierCode v1.1 Main, each model at best effort (Opus 5.5 system card: Fable 5 53.5% at xhigh) |
| FrontierCode | 46.3% | Extra highadaptive thinking, effort=xhigh | Maker's own figureVendor-reportedAnthropic ↗ | 9 Jun 2026 | Claude Code | FrontierCode Main at Fable 5 launch, metric = score (pass rate 48.8%); earlier FrontierCode version than v1.1 numbers |
| FrontierCode | 53.5% | Defaultclaude-fable-5_unknown | 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 (Diamond) | 29.3% | Extra highadaptive thinking, effort=xhigh | Maker's own figureVendor-reportedAnthropic ↗ | 9 Jun 2026 | Claude Code | FrontierCode Diamond subset, score (pass rate 30.2%); read from launch-post benchmark table image (values printed in table) |
| FrontierMath (Tiers 1-3) | 87.0% | Maxclaude-fable-5_max | Independent testIndependentEpoch AI ↗ | 9 Jun 2026 | — | FrontierMath Tiers 1-3 (v2), Epoch-run; stderr 2.0pt; data https://epoch.ai/data/benchmark_data.zip (frontiermath_tiers_1_3_v2.csv) |
| FrontierMath Tier 4 | 90.2% | Maxclaude-fable-5_max | Independent testIndependentEpoch AI ↗ | 9 Jun 2026 | — | FrontierMath Tier 4 (v2), Epoch-run; stderr 4.6pt; data https://epoch.ai/data/benchmark_data.zip (frontiermath_tier_4_v2.csv) |
| FrontierSWE v2 | 48.0% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | Proximal harness | reported as 0.48 |
| GDP.pdf | 29.8% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 9 Jun 2026 | — | no tools; read from launch-post benchmark table image (values printed in table) |
| GDPval-AA (v1) | 1932 | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 9 Jun 2026 | — | GDPval-AA (v1) Elo; higher of Mythos 5/Fable 5; read from launch-post benchmark table image (values printed in table) |
| GDPval-AA v2 | 1723 | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | read from launch-post benchmark table image (values printed in table) |
| GPQA Diamond | 93.2% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 1 Sep 2026 | — | ±1.939 stderr; $0.450/test |
| GPQA Diamond | 92.6% | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA-run evaluation (AA slug claude-fable-5) (AA marks this variant deprecated) |
| GSO | 78.4% | Extra highreasoning_effort=xhigh | Independent testIndependentGSO ↗ | 27 Sep 2026 | OpenHands | Opt@1; hack-controlled score 76.47; data https://gso-bench.github.io/assets/leaderboard.json |
| Harvey LAB-AA | 93.6% | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | criteria pass rate, AA-run |
| Humanity's Last Exam | 50.6% | Lowadaptive thinking, effort=low | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | Humanity's Last Exam without tools; vendor-reported cost/task $0.17 |
| 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 $0.4 |
| Humanity's Last Exam | 56.9% | Highadaptive thinking, effort=high | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | Humanity's Last Exam without tools; vendor-reported cost/task $0.62 |
| Humanity's Last Exam | 57.4% | Extra highadaptive thinking, effort=xhigh | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | Humanity's Last Exam without tools; vendor-reported cost/task $0.91 |
| Humanity's Last Exam | 55.5% | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA-run evaluation (AA slug claude-fable-5) (AA marks this variant deprecated) |
| Humanity's Last Exam | 57.8% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | Humanity's Last Exam without tools; vendor-reported cost/task $1.7 |
| Humanity's Last Exam (with tools) | 59.6% | Lowadaptive thinking, effort=low | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | Humanity's Last Exam with tools; vendor-reported cost/task $0.61 |
| Humanity's Last Exam (with tools) | 61.4% | Mediumadaptive thinking, effort=med | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | Humanity's Last Exam with tools; vendor-reported cost/task $1.01 |
| Humanity's Last Exam (with tools) | 63.2% | Highadaptive thinking, effort=high | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | Humanity's Last Exam with tools; vendor-reported cost/task $1.42 |
| Humanity's Last Exam (with tools) | 63.5% | Extra highadaptive thinking, effort=xhigh | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | Humanity's Last Exam with tools; vendor-reported cost/task $1.92 |
| Humanity's Last Exam (with tools) | 63.8% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | Humanity's Last Exam with tools; vendor-reported cost/task $3.44 |
| LegalBench (Vals) | 88.6% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 29 Sep 2026 | — | vals id anthropic/claude-fable-5; rank 1/149; ±0.331 stderr; $0.030639/test |
| LiveCodeBench | 89.8% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 1 Sep 2026 | — | ±0.892 stderr; $0.428/test |
| LLM Creative Story-Writing Benchmark (Lech Mazur) | 2.9 | HighClaude Fable 5 (high) | Independent testIndependentLech Mazur (LLM Creative Story-Writing Benchmark) ↗ | 28 Sep 2026 | — | rank 7/56; Thurstone comparison score (centered at 0); est. win chance 86%; 95% bootstrap 2.803 to 2.946; incomplete story set (see README coverage note) |
| LMArena Code Arena (WebDev) | 1626 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | Code Arena | WebDev overall (agentic web-dev, raw); rank 17 (CI rank 14-24); 95% CI 1619-1632; 13952 votes |
| LMArena Search Arena | 1230 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 24 Aug 2026 | — | rank 5 (rank range 3-7); 95% CI 1222.3-1238.1; 41795 votes |
| LMArena Text - Coding category | 1552 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | text arena coding category, style control on; rank 2 (CI rank 1-13); 95% CI 1545-1559; 9946 votes |
| LMArena Text - Creative Writing | 1503 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 3 (rank range 1-6); 95% CI 1495.3-1510.7; 8072 votes; style-controlled |
| LMArena Text - Expert | 1549 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 1 (rank range 1-13); 95% CI 1539.3-1558.7; 4217 votes; style-controlled |
| LMArena Text - Hard Prompts | 1531 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 4 (rank range 2-7); 95% CI 1525.5-1535.7; 24800 votes; style-controlled |
| LMArena Text - Instruction Following | 1510 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 4 (rank range 1-7); 95% CI 1503.5-1515.9; 13415 votes; style-controlled |
| LMArena Text - Longer Query | 1523 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 4 (rank range 2-7); 95% CI 1517.2-1528.8; 17723 votes; style-controlled |
| LMArena Text - Multi-Turn | 1518 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 4 (rank range 2-10); 95% CI 1509.4-1525.9; 6231 votes; style-controlled |
| LMArena Text - Non-English | 1494 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 4 (rank range 2-12); 95% CI 1488.7-1499.2; 21613 votes; style-controlled |
| LMArena Text - Occupational: Business, Management & Finance | 1502 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 5 (rank range 1-15); 95% CI 1494.4-1509.5; 7646 votes; style-controlled |
| LMArena Text - Occupational: Legal & Government | 1512 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 5 (rank range 1-28); 95% CI 1500.4-1522.9; 3185 votes; style-controlled |
| LMArena Text - Occupational: Writing, Literature & Language | 1509 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | rank 3 (rank range 1-7); 95% CI 1502.0-1515.6; 10571 votes; style-controlled |
| LMArena Text (overall) | 1505 | Highclaude-fable-5-high | Independent testIndependentLMArena ↗ | 30 Sep 2026 | — | text overall, style control on; rank 3 (CI rank 2-9); 95% CI 1500-1509; 37900 votes |
| MCP Atlas | 83.3% | DefaultClaude Fable 5 | Independent testIndependentScale AI SEAL ↗ | 1 Oct 2026 | — | rank 2 (Scale rank accounts for CI); ±2.25; entry added 2026-06-09 |
| OSWorld 2.0 | 72.9% | 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) | 36.1% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | strict pass rate; August 2026 task release |
| OSWorld-Verified | 85.0% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 9 Jun 2026 | — | read from launch-post benchmark table image (values printed in table) |
| PRBench Finance (Scale) | 53.9% | Defaultclaude-fable 5 | Independent testIndependentScale AI (SEAL) ↗ | 28 Jul 2026 | — | Scale rank 3; ±0.14 CI |
| PRBench Legal (Scale) | 52.6% | Defaultclaude-fable 5 | Independent testIndependentScale AI (SEAL) ↗ | 28 Jul 2026 | — | Scale rank 3; ±0.54 CI |
| ProgramBench (fully resolved) | 2.0% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 29 Sep 2026 | mini-SWE-agent | ±0.992 stderr; $75.676/test; strict fully-resolved rate |
| SciCode | 61.0% | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA-run evaluation (AA slug claude-fable-5) (AA marks this variant deprecated) |
| SimpleBench | 81.9% | DefaultClaude Fable | Independent testIndependentSimpleBench ↗ | 10 Jun 2026 | — | AVG@5, temp 0.7; rank 6th; human baseline 83.7%; data https://simple-bench.com/static/js/leaderboard-data.js |
| SimpleQA Verified | 70.7% | Extra highclaude-fable-5_xhigh | Independent testIndependentEpoch AI ↗ | 10 Aug 2026 | — | Epoch-run (no tools); ±1.44 stderr |
| SWE Atlas - Codebase QnA | 39.0% | Extra highFable-5 (Claude Code) xHigh* | Independent testIndependentScale AI SEAL ↗ | 1 Oct 2026 | Claude Code | rank 5 (Scale rank accounts for CI); ±5; entry added 2026-03-02; * = Scale footnote (see page) |
| SWE Atlas - Refactoring | 54.8% | Extra highFable-5 (Claude Code) xHigh | Independent testIndependentScale AI SEAL ↗ | 1 Oct 2026 | Claude Code | rank 1 (Scale rank accounts for CI); ±6.76; entry added 2026-06-11 |
| SWE Atlas - Test Writing | 55.6% | Extra highFable-5 (Claude Code) xHigh* | Independent testIndependentScale AI SEAL ↗ | 1 Oct 2026 | Claude Code | rank 1 (Scale rank accounts for CI); ±5.8; entry added 2026-06-11; * = Scale footnote (see page) |
| SWE-bench Multilingual | 86.6% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | |
| SWE-bench Multimodal | 54.1% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 24 Jul 2026 | — | |
| SWE-Bench Pro (public, v1) | 80.0% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | — | |
| SWE-bench Verified | 95.0% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 1 Sep 2026 | bash-only (single bash tool) agent | ±0.976 stderr; $2.047/test; Vals archived SWE-bench Verified on 2026-09-01 (saturated) |
| SWE-bench Verified | 95.0% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 9 Jun 2026 | — | |
| SWE-rebench | 64.5% | HighFable 5 [high] | Independent testIndependentSWE-rebench (Nebius) ↗ | 1 Oct 2026 | SWE-rebench standard scaffold | time window 2026-05-15..2026-07-01 (111 problems, 65 repos); ±1.41; pass@5 78.4%; $4.40/problem |
| tau2-bench | 98.5% | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA-run evaluation (AA slug claude-fable-5) (AA marks this variant deprecated) |
| Terminal-Bench 2.1 | 84.3% | Highadaptive thinking, effort=high | Maker's own figureVendor-reportedAnthropic ↗ | 9 Jun 2026 | mini-SWE-agent (GKE) | 20.9% of trials hit a safety refusal and fell back to Opus 4.8 |
| Terminal-Bench 2.1 | 84.6% | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA-run evaluation (AA slug claude-fable-5) (AA marks this variant deprecated) |
| Terminal-Bench 2.1 | 80.5% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 27 Sep 2026 | Terminus 2 | ±1.35 stderr; $1.429/test |
| Terminal-Bench 2.1 | 83.8% | Default | Independent testIndependentSnorkel AI / Terminal-Bench ↗ | 1 Oct 2026 | Claude Code | TB 2.1 (89 tasks, archived); effort not listed |
| Terminal-Bench 2.1 | 80.4% | Default | Independent testIndependentSnorkel AI / Terminal-Bench ↗ | 1 Oct 2026 | Terminus 2 | TB 2.1 (89 tasks, archived); effort not listed |
| Terminal-Bench 3.0 | 34.1% | Max | Independent testIndependentSnorkel AI / Terminal-Bench ↗ | 1 Oct 2026 | Claude Code | TB 3.0 v0.1 (74 tasks); superseded by TB 4.0; tokens 3.6B, run cost $6.5k |
| Terminal-Bench 4.0 | 44.5% | Max | Independent testIndependentTerminal-Bench ↗ | 1 Oct 2026 | Claude Code | TB 4.0.0 (66 tasks); ±3.85 95% CI; 330 trials; total run cost $7265; model release 2026-06-09 |
| Terminal-Bench 4.0 | 42.4% | MaxClaude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback) | Independent testIndependentArtificial Analysis ↗ | 1 Oct 2026 | — | AA-run evaluation (AA slug claude-fable-5) (AA marks this variant deprecated) |
| Terminal-Bench 4.0 | 41.4% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 29 Sep 2026 | mini-SWE-agent | ±1.336 stderr; $30.340/test |
| Terminal-Bench 4.0 | 42.0% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | Claude Code (--bare) | public leaderboard: 44.5% |
| Terminal-Bench-Science 0.1 | 12.3% | Lowadaptive thinking, effort=low | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | Claude Code (--bare) | Terminal-Bench-Science 0.1 (70 tasks); vendor-reported cost/task $17.1 |
| Terminal-Bench-Science 0.1 | 21.4% | Mediumadaptive thinking, effort=med | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | Claude Code (--bare) | Terminal-Bench-Science 0.1 (70 tasks); vendor-reported cost/task $25.0 |
| Terminal-Bench-Science 0.1 | 25.0% | Highadaptive thinking, effort=high | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | Claude Code (--bare) | Terminal-Bench-Science 0.1 (70 tasks); vendor-reported cost/task $34.3 |
| Terminal-Bench-Science 0.1 | 23.4% | Extra highadaptive thinking, effort=xhigh | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | Claude Code (--bare) | Terminal-Bench-Science 0.1 (70 tasks); vendor-reported cost/task $36.0 |
| Terminal-Bench-Science 0.1 | 24.7% | Maxadaptive thinking, effort=max | Maker's own figureVendor-reportedAnthropic ↗ | 1 Sep 2026 | Claude Code (--bare) | Terminal-Bench-Science 0.1 (70 tasks); vendor-reported cost/task $44.1 |
| Vals Code Migration | 55.1% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 29 Sep 2026 | — | ±4.605 stderr; $112.098/test |
| Vals CorpFin v2 | 71.8% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 12 Aug 2026 | — | vals id anthropic/claude-fable-5; rank 2/134; ±0.883 stderr; $1.737573/test |
| Vals Index | 61.4% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 30 Sep 2026 | — | ±0.999 stderr; $29.585/test |
| Vals Legal Research Bench | 49.5% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 29 Sep 2026 | — | vals id anthropic/claude-fable-5; rank 6/72; ±3.475 stderr; $9.791426/test |
| Vals Public Benefits Bench v1.1 | 70.4% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 29 Sep 2026 | — | vals id anthropic/claude-fable-5; rank 4/45; ±1.187 stderr; $4.585446/test |
| Vals TaxEval v2 | 76.9% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 1 Sep 2026 | — | vals id anthropic/claude-fable-5; rank 5/145; ±0.821 stderr; $0.220133/test |
| Vals Vibe Code Bench | 90.3% | Maxeffort=max | Independent testIndependentVals.ai ↗ | 29 Sep 2026 | OpenHands | ±2.097 stderr; $41.715/test |