THE FORGE RANKINGS·2026-08-30

Claude Fable 5

Anthropic
3
RANK of 37 ranked
FERROX INDEX94.8 weighted mean of category positions
GRADEA+ top of the measured field
EVIDENCE11 / 32 benchmarks measured
QUALITYTHIN Correlated evidence in: preference. Those category scores rest on benchmarks from a single family.

Not separable from GPT-5.6 Sol and Kimi K3 at the measured 1.78 point band. The order is printed; the gap is not claimed.

Compared to what

Each rail is one classification. Every ranked model measured on it is a tick; this model is the orange marker. A hatched rail means this model has no measurement in that classification and its weight was redistributed.

Agents

8of 37 measured
48.4
13.1median 38.963.0

Coding

5of 37 measured
60.1
22.4median 50.070.7

Reasoning

5of 37 measured
90.2
18.1median 62.893.0

Preference

1of 15 measured
82.5
53.1median 71.882.5

Every measurement, and the field behind it

One card per benchmark this model has been measured on. The rail shows where it sits against every other ranked model measured on the same benchmark. Where the evaluator keys on a harness, the harness is named, because the same model scores differently under a different one.

ALE-Bench

3/35
2041 best 2177 · GPT-5.6 Sol
ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0 high

APEX Agents

1/28
45.0% best 45.0% · Claude Fable 5
APEX CC-BY-4.0

LMArena Agent

2/21
0.117 best 0.127 · Claude Opus 5
LMArena CC-BY-4.0

LMArena Text (style-controlled)

1/20
1507 best 1507 · Claude Fable 5
LMArena CC-BY-4.0

FrontierCode

1/17
53.5% best 53.5% · Claude Fable 5
cognition.com CC-BY-4.0 claude-code xhigh

LMArena WebDev

1/16
1626 best 1626 · Claude Fable 5
LMArena CC-BY-4.0

Terminal-Bench 4.0

2/10
44.5% best 51.8% · Claude Opus 5
Terminal-Bench Apache-2.0 claude-code max

Not measured

21/32
  • FrontierMath
  • SWE-bench Verified (Epoch's own run)
  • Humanity's Last Exam
  • Terminal-Bench 2
  • METR time horizons
  • Aider Polyglot
  • OSWorld
  • Cybench
  • BrowseComp
  • IMOAnswerBench
  • SWE-bench Multilingual
  • AIME
  • CyberGym
  • HMMT
  • MCP-Atlas
  • SWE-bench Pro
  • Tau2-Bench Airline
  • Tau2-Bench Banking
  • Tau2-Bench Retail
  • Tau2-Bench Telecom
  • Tool-Decathlon
A missing benchmark is not a zero and is never scored as one.

Check it yourself

Every number on this page, with who measured it, the interval they published, the harness it was run under, and where to go and read it. Nothing here was measured by Ferrox Labs.

BenchmarkClassPublishedInterval (normalised) Measured byLicenceHarness / effort
APEX Agents agents 45.0% 37.0 to 53.0 APEX CC-BY-4.0 source
LMArena Agent agents 0.117 50.2 to 61.1 LMArena CC-BY-4.0 source
Terminal-Bench 4.0 agents 44.5% 40.7 to 48.4 Terminal-Bench Apache-2.0 claude-code / max source
ALE-Bench coding 2041 not published ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0 / high source
DeepSWE coding 68.6% 67.5 to 69.7 deepswe.datacurve.ai CC-BY-4.0 mini-swe-agent / high source
FrontierCode coding 53.5% not published cognition.com CC-BY-4.0 claude-code / xhigh source
LMArena Text (style-controlled) preference 1507 86.1 to 87.5 LMArena CC-BY-4.0 source
LMArena WebDev preference 1626 77.3 to 79.3 LMArena CC-BY-4.0 source
ARC-AGI-2 reasoning 87.5% not published ARC Prize CC-BY-4.0 / high source
GPQA Diamond (Epoch's own run) reasoning 83.3% 78.1 to 88.5 Epoch AI CC-BY-4.0 / high source
OTIS Mock AIME 2024-2025 reasoning 99.7% 99.2 to 100.0 Epoch AI CC-BY-4.0 / max source

Configurations rolled up: 13. Rule: median observed configuration per benchmark (lower median, always a real measurement). Harnesses seen: claude-code, mini-swe-agent.

Snapshot 2026-08-30-62e4e85f043b, manifest dc46131315046f1e. Grades are positional: position in the measured field, as a percentile of rank among ranked entries, n=37.