Moonshot AIKimi K2.6
Kimi K2.6 is a Moonshot AI open-weights model, released 20 Apr 2026. Its best configuration (thinking reasoning effort) ranks #132 of 785 on the BenchLeader Index at 59.4 ±4.1, in the upper half. The ± is the point: 226 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (74) and lowest in reasoning (54). At $1.71 per million tokens blended it is pricier than most ranked models. Output speed of 64 tokens per second puts it slower than most, with a first answer in 72.2 s. It has been measured at 3 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 20 Apr 2026.stale: no new result in six months
- Blended price
- $1.71/M
- $0.950 in · $4.00 out
- Output speed
- 64 tok/s
- measured by Artificial Analysis
- First answer
- 72 s
- first token 0.81 s
- Context
- 262k
- Full answer
- 80 s
- median, reasoning included
- Cost per run
- $0.800
- one full Intelligence Index run
- Released
- 20 Apr 2026
- Overall index59
- Reasoning54
- Agents & tools58
- Maths57
- Knowledge64
- Instruction following74
- Multimodal62
- Long context66
- Composite60
Versions
Moonshot AI has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 53.3 | #293 | 59 tok/s | 2.73 s | $0.0016 | Agents & tools 67 · Composite 56 · Instruction following 47 · Knowledge 57 · Long context 60 · Reasoning 43 |
| thinkingbest | 59.4 | #132 | 64 tok/s | 72 s | $0.0016 | Agents & tools 58 · Composite 60 · Instruction following 74 · Knowledge 64 · Long context 66 · Maths 57 · Multimodal 62 · Reasoning 54 |
| not stated | 55.4 | #232 | 63 tok/s | 0.81 s | $0.0016 | Agents & tools 34 · Coding 59 · Human preference 60 · Knowledge 56 · Maths 56 · Multimodal 62 · Reasoning 65 |
How its index has moved
8 Sept 2026 to 21 Oct 2026- thinking
- not stated
- no reasoning
The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| GPQA Diamondeffort not stated | – | – | 90.8%#39 | Epoch AI Benchmarking Hub | |
| Arena Hard Promptseffort not stated | – | – | 1485#57 | Arena (LMArena) | |
| LiveBench 1.0 Reasoningnot in index | – | 79.4%#54 | – | LiveBench | |
| GPQA Diamond (AA)not in index | 78.8%#207 | 91.1%#50 | – | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 19.6%#219 | 37.5%#105 | – | Artificial Analysis | |
| GPQA Diamond (Vals)not in indexeffort not stated | – | – | 89.1%#30 | Vals AI | |
| CritPt | 1.4%#211 | 8.0%#138 | – | Artificial Analysis | |
| Chess Puzzlesnot in indexeffort not stated | – | – | 20.0%#92 | Epoch AI Benchmarking Hub | |
| Mystery Game Puzzlesnot in index | 12.0%#103 | – | 18.0%#69 | Epoch AI Benchmarking Hub | |
| LMCAnot in indexeffort not stated | – | – | 37.3%#119 | Epoch AI Benchmarking Hub | |
| DTBenchnot in indexeffort not stated | – | – | 90.9%#76 | Epoch AI Benchmarking Hub |
Coding
| Benchmark | thinking | not stated | Source | Trend |
|---|---|---|---|---|
| SWE-bench Verified (Epoch)effort not stated | – | 76.7%#9 | Epoch AI Benchmarking Hub | |
| SciCodeeffort not stated | – | 53.5%#75 | SciCode | |
| WeirdMLeffort not stated | – | 55.9%#58 | WeirdML | |
| Arena Codingeffort not stated | – | 1516#44 | Arena (LMArena) | |
| Arena WebDeveffort not stated | – | 1509#57 | Arena (LMArena) | |
| LiveBench 1.0 Codingnot in index | 78.6%#30 | – | LiveBench | |
| SciCode (AA)not in index | 51.5%#96 | – | Artificial Analysis | |
| LiveCodeBencheffort not stated | – | 86.8%#19 | Vals AI | |
| SWE-bench (Vals)not in indexeffort not stated | – | 76.2%#39 | Vals AI | |
| Code Migrationnot in indexeffort not stated | – | 27.8%#43 | Vals AI | |
| ProgramBenchnot in indexeffort not stated | – | 0.0%#32 | Vals AI | |
| Vibe Code Bench v1.1not in indexeffort not stated | – | 37.9%#69 | Vals AI | |
| ALE-Benchnot in indexeffort not stated | – | 1092.7#51 | Epoch AI Benchmarking Hub | |
| Terminal-Bench 4.0 (AA)not in index | 0.5%#139 | – | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 65.9%#80 | – | Artificial Analysis |
Agents & tools
Maths
| Benchmark | thinking | not stated | Source | Trend |
|---|---|---|---|---|
| FrontierMath Tiers 1–3effort not stated | – | 57.2%#45 | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4effort not stated | – | 25.6%#49 | Epoch AI Benchmarking Hub | |
| OTIS Mock AIMEeffort not stated | – | 96.1%#40 | Epoch AI Benchmarking Hub | |
| ProofBencheffort not stated | – | 16.0%#56 | Vals AI | |
| LiveBench Mathematicsnot in index | 84.3%#57 | – | LiveBench | |
| AIME 2026possible contamination | 95.8%#13 | – | MathArena | |
| HMMT February 2026possible contamination | 94.7%#8 | – | MathArena | |
| USAMO 2026not in indexpossible contamination | 51.2%#5 | – | MathArena | |
| MathArena Apexpossible contamination | 24.0%#13 | – | MathArena | |
| Arena Mathseffort not stated | – | 1479#39 | Arena (LMArena) |
Knowledge
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| SimpleQA Verifiedeffort not stated | – | – | 34.9%#52 | Epoch AI Benchmarking Hub | |
| LiveBench 1.0 Data Analysisnot in index | – | 65.1%#57 | – | LiveBench | |
| AA-Omniscience | -9.2#186 | 5.3#106 | – | Artificial Analysis | |
| MMLU-Proeffort not stated | – | – | 87.6%#25 | Vals AI | |
| LegalBencheffort not stated | – | – | 84.7%#30 | Vals AI | |
| CorpFineffort not stated | – | – | 66.7%#16 | Vals AI | |
| TaxEvaleffort not stated | – | – | 74.7%#32 | Vals AI | |
| CaseLaw v2not in indexeffort not stated | – | – | 61.2%#22 | Vals AI | |
| Excel Modeling Benchmarknot in indexeffort not stated | – | – | 57.9%#39 | Vals AI | |
| MedCodenot in indexeffort not stated | – | – | 40.1%#65 | Vals AI | |
| MedScribenot in indexeffort not stated | – | – | 78.2%#64 | Vals AI | |
| Public Benefits Bench v1.1not in indexeffort not stated | – | – | 56.6%#37 | Vals AI | |
| EBR-benchnot in indexeffort not stated | – | – | 2.4%#26 | Epoch AI Benchmarking Hub | |
| GDP.pdfnot in indexeffort not stated | – | – | 12.0%#48 | Epoch AI Benchmarking Hub | |
| AA-Omniscience: accuracynot in index | 24.3%#247 | 32.6%#169 | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 55.8%#93 | 59.5%#80 | – | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | 75.1%#47 | – | LiveBench | |
| LiveBench Instruction Followingnot in index | – | 64.4%#47 | – | LiveBench | |
| IFBench | 44.3%#215 | 76.0%#23 | – | Artificial Analysis | |
| Arena Instruction Followingnot in indexeffort not stated | – | – | 1455#57 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1461#59 | Arena (LMArena) | |
| EQ-Bench 4 | 1202#16 | EQ-Bench | |
| Arena Creative Writingnot in index | 1432#70 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1461#75 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1477#51 | Arena (LMArena) |
Multimodal
| Benchmark | thinking | not stated | Source | Trend |
|---|---|---|---|---|
| Arena Visioneffort not stated | – | 1282#34 | Arena (LMArena) | |
| MMMU-Pro | 79.4%#61 | – | Artificial Analysis | |
| MMMU-Pro (Vals)not in indexeffort not stated | – | 86.3%#21 | Vals AI | |
| MortgageTaxnot in indexeffort not stated | – | 65.8%#38 | Vals AI | |
| SAGEnot in indexeffort not stated | – | 50.2%#17 | Vals AI | |
| Blueprint-Bench 2not in indexeffort not stated | – | 3.9%#28 | Epoch AI Benchmarking Hub |
Long context
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| AA-LCR | 69.7%#222 | 81.0%#61 | – | Artificial Analysis | |
| Arena Documentnot in indexeffort not stated | – | – | 1450#32 | Arena (LMArena) | |
| MLCRnot in index | – | 11.1%#104 | – | Artificial Analysis |
Composite
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in indexeffort not stated | – | – | 150.9#54 | Epoch AI Benchmarking Hub | |
| LiveBench 1.0not in index | – | 70.5%#54 | – | LiveBench | |
| AA Intelligence Index v4.3.2 | 23.6#185 | 27.0#147 | – | Artificial Analysis | |
| Vals Multimodal Indexnot in indexeffort not stated | – | – | 56.4%#19 | Vals AI |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| AtlasCloud | 94 tok/s | 1.01 s | $0.950 | $4.00 | 262k | int4 |
| CoreWeave | 87 tok/s | 0.71 s | $0.650 | $3.41 | 262k | fp4 |
| DeepInfra | 81 tok/s | 0.96 s | $0.750 | $3.50 | 262k | fp4 |
| Venice | 75 tok/s | 0.61 s | $0.750 | $3.50 | 256k | int4 |
| Baidu Qianfan | 74 tok/s | 0.81 s | $0.449 | $1.89 | 262k | fp4 |
| Inceptron | 74 tok/s | 0.78 s | $0.603 | $3.20 | 262k | int4 |
| Parasail | 66 tok/s | 0.80 s | $0.750 | $3.50 | 262k | int4 |
| Crusoe | 59 tok/s | 0.30 s | $0.700 | $3.50 | 262k | bf16 |
| Chutes | 55 tok/s | 1.79 s | $0.450 | $1.95 | 262k | int4 |
| Moonshot AI | 52 tok/s | 1.74 s | $0.950 | $4.00 | 262k | int4 |
| Phala | 49 tok/s | 1.39 s | $1.09 | $4.60 | 262k | – |
| DigitalOcean | 47 tok/s | 0.51 s | $0.570 | $2.40 | 262k | – |
| StreamLake | 45 tok/s | 1.36 s | $0.599 | $2.52 | 256k | fp8 |
| NovitaAI | 32 tok/s | 1.80 s | $0.800 | $3.40 | 262k | – |
| Cloudflare | 30 tok/s | 1.24 s | $0.950 | $4.00 | 262k | – |
| SiliconFlow | 23 tok/s | 1.35 s | $0.770 | $3.40 | 262k | fp8 |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.160 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0016 | $0.0013 | 1.3 min |
| Summarize a 30-page report | 12,000 / 600 | $0.014 | $0.0067 | 1.4 min |
| Code edit | 6,000 / 1,500 | $0.012 | $0.0081 | 1.6 min |
| Agentic coding session | 60,000 / 4,000 | $0.073 | $0.037 | 2.2 min |
| Structured extraction | 2,000 / 200 | $0.0027 | $0.0015 | 1.3 min |
More from Moonshot AI
See also
Questions people ask
- What are Kimi K2.6's benchmark scores?
- As of 24 Oct 2026, Kimi K2.6 (thinking effort) scores 59.4 ±4.1 on the BenchLeader Index, strongest in instruction following (74) and weakest in reasoning (54). That rests on 11 independent benchmark results, and every score on this page links to its publisher.
- How does Kimi K2.6 rank?
- Kimi K2.6 (thinking effort) is #132 of 785 ranked configurations on 24 Oct 2026. 226 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Kimi K2.6 cost per 1M tokens?
- As of 24 Oct 2026, Kimi K2.6 costs $0.950 per 1M input tokens and $4.00 per 1M output tokens, $1.71 blended (three input tokens per output token).
- When was Kimi K2.6 released?
- Moonshot AI released Kimi K2.6 on 20 Apr 2026. BenchLeader has results for it at 3 reasoning-effort settings.
Data as of 24 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “Kimi K2.6: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/kimi-k2-6.