THE FORGE RANKINGS·2026-08-30

Grok 4.6

xAI
5
RANK of 37 ranked
FERROX INDEX92.8 weighted mean of category positions
GRADEA front rank
EVIDENCE9 / 32 benchmarks measured
QUALITYFULL every category rests on independent benchmarks

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

27of 37 measured
32.8
13.1median 38.963.0

Coding

16of 37 measured
52.1
22.4median 50.070.7

Reasoning

6of 37 measured
84.1
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

9/35
1508 best 2177 · GPT-5.6 Sol
ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0 xhigh

APEX Agents

4/28
41.2% best 45.0% · Claude Fable 5
APEX CC-BY-4.0

LMArena Agent

9/21
0.061 best 0.127 · Claude Opus 5
LMArena CC-BY-4.0

FrontierCode

3/17
48.0% best 53.5% · Claude Fable 5
cognition.com CC-BY-4.0 grok-build high

Terminal-Bench 4.0

7/10
20.3% best 51.8% · Claude Opus 5
Terminal-Bench Apache-2.0 grok-build none

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 41.2% 33.2 to 49.2 APEX CC-BY-4.0 source
LMArena Agent agents 0.061 32.2 to 41.9 LMArena CC-BY-4.0 source
Terminal-Bench 4.0 agents 20.3% 17.2 to 23.4 Terminal-Bench Apache-2.0 grok-build / none source
ALE-Bench coding 1508 not published ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0 / xhigh source
DeepSWE coding 65.2% 63.6 to 66.7 deepswe.datacurve.ai CC-BY-4.0 mini-swe-agent / high source
FrontierCode coding 48.0% not published cognition.com CC-BY-4.0 grok-build / high source
ARC-AGI-2 reasoning 61.3% not published ARC Prize CC-BY-4.0 / medium source
GPQA Diamond (Epoch's own run) reasoning 93.2% 90.2 to 96.2 Epoch AI CC-BY-4.0 / xhigh source
OTIS Mock AIME 2024-2025 reasoning 97.8% 95.8 to 99.7 Epoch AI CC-BY-4.0 / high source

Configurations rolled up: 11. Rule: median observed configuration per benchmark (lower median, always a real measurement). Harnesses seen: grok-build, 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.