THE FORGE RANKINGS·2026-08-30

GPT-5.6 Sol

OpenAI
2
RANK of 37 ranked
FERROX INDEX95.5 weighted mean of category positions
GRADEA+ top of the measured field
EVIDENCE9 / 32 benchmarks measured
QUALITYFULL every category rests on independent benchmarks

Not separable from Claude Fable 5 at the measured 1.78 point band. The order is printed; the gap is not claimed.

No measurement in Preference. That weight was redistributed, which is the same as assuming this model would have scored its own average there. It is an assumption, not a measurement.

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

17of 37 measured
41.9
13.1median 38.963.0

Coding

6of 37 measured
59.7
22.4median 50.070.7

Reasoning

4of 37 measured
90.3
18.1median 62.893.0

Preference

no qualifying measurement
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

1/35
2177 best 2177 · GPT-5.6 Sol
ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0 max

LMArena Agent

3/21
0.102 best 0.127 · Claude Opus 5
LMArena CC-BY-4.0

FrontierCode

4/17
47.5% best 53.5% · Claude Fable 5
cognition.com CC-BY-4.0 codex max

Terminal-Bench 4.0

4/10
37.3% best 51.8% · Claude Opus 5
Terminal-Bench Apache-2.0 codex max

Not measured

23/32
  • FrontierMath
  • SWE-bench Verified (Epoch's own run)
  • LMArena Text (style-controlled)
  • Humanity's Last Exam
  • Terminal-Bench 2
  • LMArena WebDev
  • 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 37.7% 30.1 to 45.3 APEX CC-BY-4.0 / xhigh source
LMArena Agent agents 0.102 45.8 to 55.7 LMArena CC-BY-4.0 source
Terminal-Bench 4.0 agents 37.3% 33.5 to 41.0 Terminal-Bench Apache-2.0 codex / max source
ALE-Bench coding 2177 not published ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0 / max source
DeepSWE coding 69.4% 68.0 to 70.8 deepswe.datacurve.ai CC-BY-4.0 mini-swe-agent / high source
FrontierCode coding 47.5% not published cognition.com CC-BY-4.0 codex / max source
ARC-AGI-2 reasoning 85.4% not published ARC Prize CC-BY-4.0 / high source
GPQA Diamond (Epoch's own run) reasoning 89.9% 85.7 to 94.1 Epoch AI CC-BY-4.0 / low source
OTIS Mock AIME 2024-2025 reasoning 95.6% 89.5 to 100.0 Epoch AI CC-BY-4.0 / low source

Configurations rolled up: 13. Rule: median observed configuration per benchmark (lower median, always a real measurement). Harnesses seen: codex, 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.