Skip to content
Leaderboard / Moonshot AI

Kimi K2.6

Moonshot AI OPEN Modified MIT
Verdict

Kimi K2.6 is a strong all-rounder: #57 of 406 by SIscope score. Strongest at chat quality, weaker at reasoning & knowledge. Mid-range on price. Weights are open, so you can run it yourself.

SIscope score
76
#57 of 406
Human-vote rank
#45
of 419 · score 1456
Price per 1M tokens, blended
$1.34
$0.65 in · $3.41 out
Context window
262K
262,144 tokens

Shape

Each spoke is a category, 0 at the centre, 100 at the edge.

CodingAgentsno data yetReasoningMathChatInstructions

Category scores

Percentiles: 90 means better than about 90% of tested models.

Coding74
Agents & toolsNo data yet–
Reasoning & knowledge73
Math79
Chat quality90
Following instructions88

What goes into each category · Compare with Gemini 4 Carbon

Human-vote score over time

Text arena, recent updates.

145014551460Oct 20: 1455.5 (rank #44)Oct 22: 1455.3 (rank #44)Oct 28: 1455.6 (rank #45)Oct 20Oct 28

Coding votes over time

WebDev arena, recent updates.

150515101515Oct 16: 1509.6 (rank #56)Oct 18: 1509.6 (rank #57)Oct 19: 1509.5 (rank #59)Oct 22: 1509.4 (rank #59)Oct 23: 1509.4 (rank #60)Oct 27: 1509.3 (rank #61)Oct 28: 1509.2 (rank #62)Oct 16Oct 28

Independent test runs

Run by Epoch AI on the same harness for every model. Bars show the share of questions answered correctly.

GPQA Diamond91%
PhD-level science multiple choice questions
Better than 88% of 333 tested models. Run May 1, 2026.
SimpleQA Verified35%
Short factual questions; rewards not making things up
Better than 41% of 100 tested models. Run Aug 10, 2026.
OTIS Mock AIME 2024-2596%
Competition maths problems in the style of AIME
Better than 85% of 313 tested models. Run May 2, 2026.
FrontierMath Tiers 1-357%
Very hard, unpublished research-style maths problems
Better than 61% of 122 tested models. Run Jun 10, 2026.
SWE-bench Verified77%
Fixing real GitHub issues in Python repositories
Better than 71% of 35 tested models. Run May 8, 2026.

Human-vote arenas

People compare two anonymous answers and pick the better one. Scores work like chess ratings.

ArenaRankScoreRange (95%)VotesUpdated
Text arena
Everyday chat, blind human votes
#45 of 41914561451–146039,978Oct 28, 2026
WebDev arena
Building web apps, blind human votes
#62 of 15215091502–151710,119Oct 28, 2026
Vision arena
Questions about images, blind human votes
#34 of 16712821275–128916,368Oct 19, 2026
Agent arena
Real multi-step agent sessions
Not on this board yet

Source: Arena (formerly LMArena) leaderboard dataset, CC BY 4.0. Fetched Nov 4, 2026.

Price

Input
$0.65
per 1M tokens
Output
$3.41
per 1M tokens
Blended
$1.34
3 in : 1 out, per 1M
Context window
262K
max tokens per request
Speed
Coming soon
our own tests

List price from the OpenRouter models API (moonshotai/kimi-k2.6), Nov 4, 2026. Providers can charge differently.