THE FORGE RANKINGS·2026-08-30

Claude 4.5 Opus

Anthropic
24
RANK of 37 ranked
FERROX INDEX61.4 weighted mean of category positions
GRADEC below the field median
EVIDENCE12 / 32 benchmarks measured
QUALITYTHIN Correlated evidence in: preference. Those category scores rest on benchmarks from a single family.

Not separable from Kimi K2.7 Code 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

6of 37 measured
50.0
13.1median 38.963.0

Coding

15of 37 measured
53.0
22.4median 50.070.7

Reasoning

29of 37 measured
41.2
18.1median 62.893.0

Preference

9of 15 measured
69.9
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

22/35
1025 best 2177 · GPT-5.6 Sol
ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0 16k

APEX Agents

20/28
20.7% best 45.0% · Claude Fable 5
APEX CC-BY-4.0

SWE-bench Verified (Epoch's own run)

4/21
76.7% best 83.5% · Claude Opus 4.7
Epoch AI CC-BY-4.0

LMArena Text (style-controlled)

9/20
1469 best 1507 · Claude Fable 5
LMArena CC-BY-4.0

LMArena WebDev

11/16
1468 best 1626 · Claude Fable 5
LMArena CC-BY-4.0

Not measured

20/32
  • LMArena Agent
  • DeepSWE
  • FrontierCode
  • Terminal-Bench 4.0
  • 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 20.7% 14.6 to 26.8 APEX CC-BY-4.0 source
METR time horizons agents 4.8h 58.2 to 88.2 METR - Measuring AI Ability to Complete Long Tasks CC-BY-4.0 / 16k source
Terminal-Bench 2 agents 59.1% 54.4 to 63.8 Terminal-Bench v2 Leaderboard CC-BY-4.0 Letta Code / 128k source
ALE-Bench coding 1025 not published ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0 / 16k source
SWE-bench Verified (Epoch's own run) coding 76.7% 72.9 to 80.4 Epoch AI CC-BY-4.0 source
LMArena Text (style-controlled) preference 1469 80.9 to 81.8 LMArena CC-BY-4.0 source
LMArena WebDev preference 1468 57.6 to 59.4 LMArena CC-BY-4.0 source
ARC-AGI-2 reasoning 13.9% not published ARC Prize CC-BY-4.0 / 8k source
FrontierMath reasoning 20.7% 16.0 to 25.4 Epoch AI CC-BY-4.0 source
GPQA Diamond (Epoch's own run) reasoning 85.5% 81.3 to 89.7 Epoch AI CC-BY-4.0 / 16k source
Humanity's Last Exam reasoning 14.2% 11.5 to 16.9 Humanity’s Last Exam (CAIS / Scale AI) CC-BY-4.0 source
OTIS Mock AIME 2024-2025 reasoning 81.7% 71.8 to 91.5 Epoch AI CC-BY-4.0 / 16k source

Configurations rolled up: 7. Rule: median observed configuration per benchmark (lower median, always a real measurement). Harnesses seen: droid, letta-code.

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.