THE FORGE RANKINGS·2026-08-30

Kimi K2.6

Moonshot
18
RANK of 37 ranked
FERROX INDEX65.9 weighted mean of category positions
GRADEC below the field median
EVIDENCE10 / 32 benchmarks measured
QUALITYTHIN Correlated evidence in: preference. Those category scores rest on benchmarks from a single family.

Not separable from GPT-5.2 (2025-12-11) and Gemini 3 Pro Preview and Gemini 3 Flash Preview 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

22of 37 measured
36.9
13.1median 38.963.0

Coding

10of 37 measured
53.9
22.4median 50.070.7

Reasoning

12of 37 measured
79.2
18.1median 62.893.0

Preference

8of 15 measured
71.8
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.

GPQA Diamond (Epoch's own run)

8/36
90.8% best 94.1% · Gemini 3.1 Pro Pr…
Epoch AI CC-BY-4.0

OTIS Mock AIME 2024-2025

6/36
96.1% best 99.7% · Claude Fable 5
Epoch AI CC-BY-4.0

ALE-Bench

21/35
1093 best 2177 · GPT-5.6 Sol
ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0

APEX Agents

21/28
18.9% best 45.0% · Claude Fable 5
APEX CC-BY-4.0

LMArena Agent

18/21
0.007 best 0.127 · Claude Opus 5
LMArena CC-BY-4.0

FrontierMath

5/21
39.0% best 47.6% · GPT-5.4 (2026-03-…
Epoch AI CC-BY-4.0

SWE-bench Verified (Epoch's own run)

5/21
76.7% best 83.5% · Claude Opus 4.7
Epoch AI CC-BY-4.0

LMArena Text (style-controlled)

11/20
1461 best 1507 · Claude Fable 5
LMArena CC-BY-4.0

LMArena WebDev

8/16
1509 best 1626 · Claude Fable 5
LMArena CC-BY-4.0

OSWorld

2/6
73.1% best 75.2% · MiniMax-M3
OSWorld (XLANG Lab) none stated General model 100 steps

Not measured

22/32
  • ARC-AGI-2
  • DeepSWE
  • FrontierCode
  • Humanity's Last Exam
  • Terminal-Bench 2
  • METR time horizons
  • Terminal-Bench 4.0
  • Aider Polyglot
  • 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 18.9% 13.4 to 24.4 APEX CC-BY-4.0 source
LMArena Agent agents 0.007 10.6 to 27.1 LMArena CC-BY-4.0 source
OSWorld agents 73.1% not published OSWorld (XLANG Lab) none stated General model / 100 steps source
ALE-Bench coding 1093 not published ALE-Bench (Sakana AI with AtCoder) CC-BY-4.0 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 1461 79.5 to 80.8 LMArena CC-BY-4.0 source
LMArena WebDev preference 1509 62.7 to 64.5 LMArena CC-BY-4.0 source
FrontierMath reasoning 39.0% 33.3 to 44.6 Epoch AI CC-BY-4.0 source
GPQA Diamond (Epoch's own run) reasoning 90.8% 87.4 to 94.2 Epoch AI CC-BY-4.0 source
OTIS Mock AIME 2024-2025 reasoning 96.1% 91.5 to 100.0 Epoch AI CC-BY-4.0 source

Configurations rolled up: 2. Rule: median observed configuration per benchmark (lower median, always a real measurement). Harnesses seen: general-model.

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.