LLMs TigerLive LLM benchmarks

Kimi K2.6: benchmark scores, price and ranking

Kimi K2.6 from Moonshot ranks #357 of 743 LLMs on the LLMs Tiger aggregate benchmark score (48.4), based on 27 benchmark results captured up to 2026-10-13.

Organisation
Moonshot
Type
Open-weight / local model
Open weights
Yes
Released
2026-04-20 (source: epoch)
Cheapest API price
$0.650 in / $3.41 out per million tokens (via openrouter, observed 2026-10-13T19:47Z)
Aggregate rank
#357 of 743 (score 48.4)

Relative to other models, Kimi K2.6 is strongest on Epoch ECI (#8 of 204) and weakest on EBR-Bench (#21 of 21).

Across the 23 benchmarks where at least ten models are scored, Kimi K2.6 is in the top 10% on 3 and in the top half on 15.

Its cheapest listed input price, $0.650 per million tokens, is lower than that of 43% of the 427 models with a listed price.

Kimi K2.6 benchmark results

Kimi K2.6 benchmark scores, ranked against all models measured on each benchmark
BenchmarkScoreRankSourceCaptured
Epoch ECI151#8 of 204Epoch AI Benchmarking Hub2026-10-13
Vending-Bench 2$6,205#1 of 19Epoch AI Benchmarking Hub2026-10-13
SWE-bench Verified76.7%#4 of 73Epoch AI Benchmarking Hub2026-10-13
GPQA Diamond90.5%#32 of 275HuggingFace GPQA Diamond leaderboard2026-10-13
OTIS Mock AIME96.1%#24 of 189Epoch AI Benchmarking Hub2026-10-13
DTBench90.9%#22 of 146Epoch AI Benchmarking Hub2026-10-13
WeirdML55.9%#21 of 109Epoch AI Benchmarking Hub2026-10-13
Chess Puzzles26.0%#30 of 122Epoch AI Benchmarking Hub2026-10-13
SciCode53.5%#28 of 106Epoch AI Benchmarking Hub2026-10-13
ALE-Bench1,093#23 of 79Epoch AI Benchmarking Hub2026-10-13
WebDev Arena1,509#22 of 67Epoch AI Benchmarking Hub2026-10-13
LMArena Document Elo1,451#6 of 18LMArena2026-10-13
Blueprint-Bench 23.9%#1 of 3Epoch AI Benchmarking Hub2026-10-13
LMCA37.3%#38 of 108Epoch AI Benchmarking Hub2026-10-13
CritPt8.0%#44 of 115Epoch AI Benchmarking Hub2026-10-13
GBA-Eval0.9%#3 of 7Epoch AI Benchmarking Hub2026-10-13
FrontierMath T1-357.2%#33 of 68Epoch AI Benchmarking Hub2026-10-13
ExploitBench18.4%#2 of 4Epoch AI Benchmarking Hub2026-10-13
SimpleQA Verified34.9%#41 of 69Epoch AI Benchmarking Hub2026-10-13
FrontierMath T425.6%#31 of 50Epoch AI Benchmarking Hub2026-10-13
Mystery Games18.0%#37 of 59Epoch AI Benchmarking Hub2026-10-13
ProofBench16.0%#30 of 46Epoch AI Benchmarking Hub2026-10-13
LMArena Vision Elo1,265#15 of 22LMArena2026-10-05
OSWorld 24.6%#6 of 8Epoch AI Benchmarking Hub2026-10-13
GDP.pdf12.0%#23 of 25Epoch AI Benchmarking Hub2026-10-13
Furniture Assembly21.7%#26 of 28Epoch AI Benchmarking Hub2026-10-13
EBR-Bench2.4%#21 of 21Epoch AI Benchmarking Hub2026-10-13

About Kimi K2.6

Kimi K2.6 is an open-weight model that can run on your own hardware from Moonshot. It was released on 2026-04-20, according to Epoch AI. The cheapest API listing we track charges $0.650 per million input tokens and $3.41 per million output tokens, through openrouter. We hold 27 benchmark results for it across 4 areas, and 3 are in the top tenth of their boards.

Where Kimi K2.6 stands

On our aggregate score, Kimi K2.6 is #357 of 743 ranked models, around the middle of the table. Its score is 48.4; the scale is centred on 50, so it sits 1.6 points below the typical model. That beats 52% of the ranked field. Among the 13 ranked Moonshot models it is #9. The next one up is kimi-k2-0711-preview at 49.1. The next one down is Kimi K2 Thinking at 47.8.

Kimi K2.6, benchmark by benchmark

On Epoch ECI, Kimi K2.6 scores 151 and ranks #8 of 204. Only 7 models score higher, led by Qwen 3.8 Max at 156, which is 5 points ahead. It covers Epoch Capabilities Index: Epoch's own aggregate of ~40 benchmarks. The board median is 127, so Kimi K2.6 is 24 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Vending-Bench 2 is where Kimi K2.6 does best: $6,205, first of 19. The next result is GLM-5.1 at $5,634, 570 points behind. It covers running a vending business for a year (final money, $). The board median is $2,377, so Kimi K2.6 is 3,828 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

On SWE-bench Verified, Kimi K2.6 scores 76.7% and ranks #4 of 73. Only 3 models score higher, led by Claude Opus 4.7 (max) at 83.5%, which is 6.8 points ahead. It covers real GitHub issues resolved. The board median is 56.4%, so Kimi K2.6 is 20.3 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

GPQA Diamond puts Kimi K2.6 at 90.5%, #32 of 275. That beats 88% of the models tested, but 5.3 points short of GPT-6 Astra (max) (95.8%). The results either side are Claude Opus 5.5 (max) (90.6%) and GPT-6 Luna (max) (90.5%). It covers expert-level science questions. The board median is 71.7%, so Kimi K2.6 is 18.8 points above it. Source: HuggingFace GPQA Diamond leaderboard, captured 2026-10-13.

On OTIS Mock AIME, Kimi K2.6 reaches 96.1%. It ranks #24 of 189, ahead of 87% of the field, and 3.9 points behind Claude Fable 5.1 (max) at 100.0%. The results either side are DeepSeek v4 Pro (max) (96.7%) and GPT-5.2 (xhigh) (96.1%). It covers competition maths (AIME 2024-25 style). The board median is 57.8%, so Kimi K2.6 is 38.3 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Kimi K2.6 manages 90.9% on DTBench, good for #22 of 146. Claude Opus 5.5 (max) leads at 98.9%, 8.0 points further up, and 85% of models sit below. The results either side are GPT-5.2 (xhigh) (90.9%) and DeepSeek v4 Pro (max) (90.7%). It covers dynamic tool-use benchmark. The board median is 68.8%, so Kimi K2.6 is 22.1 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Kimi K2.6 lands below the middle on LMArena Vision Elo: 1,265, #15 of 22, with 64% of the field scoring higher. Gemini 3.7 Flash (high) leads at 1,296; the closest result above is Gemini 3 Flash at 1,273. It covers crowd-voted image understanding. The board median is 1,279, so Kimi K2.6 is 14 points below it. Source: LMArena, captured 2026-10-05.

GDP.pdf is a weak spot. Kimi K2.6 scores 12.0%, #23 of 25, behind 88% of the models measured. The leader, GPT-6 Astra (max), reaches 34.2%. It covers economically valuable document tasks. The board median is 22.8%, so Kimi K2.6 is 10.8 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Furniture Assembly is a weak spot. Kimi K2.6 scores 21.7%, #26 of 28, behind 89% of the models measured. The leader, Claude Opus 5.5 (max), reaches 83.3%. It covers planning a physical assembly. The board median is 41.2%, so Kimi K2.6 is 19.6 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

EBR-Bench is a weak spot. Kimi K2.6 scores 2.4%, #21 of 21, behind 95% of the models measured. The leader, GPT-6 Astra (max), reaches 76.2%. It covers evidence-based reasoning. The board median is 31.4%, so Kimi K2.6 is 29.0 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

The full table above also lists the other 17 results, including boards with fewer than ten models, where a rank says little.

Strengths and weaknesses

Software engineering: strong, ahead of 77% of the field on average across 4 benchmarks. Its best board here is SWE-bench Verified (#4 of 73) and its weakest is WebDev Arena (#22 of 67). Reasoning, maths and science: solid, ahead of 60% of the field on average across 6 benchmarks. Its best board here is GPQA Diamond (#32 of 275) and its weakest is ProofBench (#30 of 46). Agents and tool use: uneven, ahead of 46% of the field on average across 2 benchmarks. Its best board here is DTBench (#22 of 146) and its weakest is GDP.pdf (#23 of 25). General knowledge and chat: solid, ahead of 54% of the field on average across 11 benchmarks. Its best board here is Epoch ECI (#8 of 204) and its weakest is EBR-Bench (#21 of 21).

Price and value

The median input price across the 427 priced models is $0.500 per million tokens, so Kimi K2.6 costs more than the typical listing. Models that score within 1.5 points or better and cost less: Qwen2.5-Coder-7B-Instruct ($0.010, 49.5), DeepSeek V4.1 Flash (high) ($0.010, 48.9), Llama-3.2-3B-Instruct ($0.020, 47.2). Similar scores cost more on Kimi K2.7 Code ($0.671) and Mistral Large 4 (high) ($0.680).

Release history

The Moonshot model released before it was Kimi K2.5 (instant) on 2026-01-27, which scores 49.1. The next one was Kimi K2.7 Code on 2026-06-12, at 47.3.

Reading these numbers

The 27 scores on this page come from 3 sources: Epoch AI Benchmarking Hub (24), LMArena (2), HuggingFace GPQA Diamond leaderboard (1). No official leaderboard supplied any of them, so treat them as third-party measurements. The leaderboard also lists 1 other reasoning-effort setting of this model, which redirect here. This page shows the highest.

Kimi K2.6 is also listed at 1 other reasoning-effort setting; this page shows the highest-effort row, which is the one the leaderboard keeps by default.

Models ranked nearby

More from Moonshot

Compare it on the live leaderboard → · How the aggregate score works