Moonshot AIKimi K2 Thinking
Kimi K2 Thinking is a Moonshot AI open-weights reasoning model, released 6 Nov 2025. Its best configuration (thinking reasoning effort) ranks #316 of 785 on the BenchLeader Index at 52.4 ±3.0, in the upper half. The ± is the point: 284 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 long context (61) and lowest in coding (49). At $1.07 per million tokens blended it is pricier than most ranked models. Output speed of 121 tokens per second puts it faster than most, with a first answer in 18.4 s. Last measured 13 Oct 2026.
- Blended price
- $1.07/M
- $0.600 in · $2.50 out
- Output speed
- 121 tok/s
- measured by Artificial Analysis
- First answer
- 18 s
- first token 1.41 s
- Context
- 262k
- Full answer
- 23 s
- median, reasoning included
- Cost per run
- –
- Released
- 6 Nov 2025
- Overall index52
- Reasoning50
- Coding49
- Agents & tools50
- Maths50
- Knowledge51
- Instruction following57
- Long context61
- Composite54
How its index has moved
9 Sept 2026 to 23 Oct 2026The 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 | thinking | Source | Trend |
|---|---|---|---|
| GPQA Diamond (AA)not in index | 83.8%#153 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 23.8%#191 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | 78.5%#70 | Vals AI | |
| Kagi LLM Benchmark | 64.4%#39 | Kagi LLM Benchmark | |
| CritPt | 2.6%#189 | Artificial Analysis | |
| CL-benchnot in index | 17.6%#15 | Epoch AI Benchmarking Hub |
Coding
| Benchmark | thinking | Source | Trend |
|---|---|---|---|
| WeirdML | 42.8%#92 | WeirdML | |
| LiveCodeBench | 63.1%#97 | Vals AI | |
| SWE-bench (Vals)not in index | 60.2%#75 | Vals AI | |
| SWE-bench Verified (bash only) | 63.4%#21 | SWE-bench | |
| SWE-bench Verified (any scaffold)not in index | 63.4%#25 | SWE-bench | |
| ALE-Benchnot in index | 597.5#98 | Epoch AI Benchmarking Hub |
Agents & tools
| Benchmark | thinking | Source | Trend |
|---|---|---|---|
| Terminal-Bench | 35.7%#53 | Terminal-Bench | |
| Terminal-Bench Hard | 31.1%#106 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 93.0%#46 | Artificial Analysis | |
| Poker Agentnot in index | 1011.6%#13 | Vals AI | |
| Terminal-Bench 2.0 (Vals)not in index | 37.1%#38 | Vals AI | |
| METR Time Horizonsnot in index | 59.2%#21 | METR |
Maths
| Benchmark | thinking | Source | Trend |
|---|---|---|---|
| AIME (Vals) | 85.4%#36 | Vals AI | |
| MGSM | 90.2%#38 | Vals AI | |
| MathArena Apexpossible contamination | 0.0%#46 | MathArena |
Knowledge
| Benchmark | thinking | Source | Trend |
|---|---|---|---|
| AA-Omniscience | -21.4#246 | Artificial Analysis | |
| MMLU-Pro | 81.1%#80 | Vals AI | |
| LegalBench | 80.2%#86 | Vals AI | |
| CorpFin | 60.6%#67 | Vals AI | |
| TaxEval | 71.7%#71 | Vals AI | |
| MedQA | 92.6%#28 | Vals AI | |
| PRBench Finance | 43.4%#24 | Scale AI SEAL | |
| PRBench Legal | 40.9%#29 | Scale AI SEAL | |
| CaseLaw v2not in index | 65.7%#10 | Vals AI | |
| AA-Omniscience: accuracynot in index | 30.9%#175 | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 24.4%#248 | Artificial Analysis |
Instruction following
| Benchmark | thinking | Source | Trend |
|---|---|---|---|
| IFBench | 68.1%#82 | Artificial Analysis | |
| MultiChallenge | 55.4%#22 | Scale AI SEAL |
Long context
| Benchmark | thinking | Source | Trend |
|---|---|---|---|
| AA-LCR | 72.0%#191 | Artificial Analysis |
Composite
| Benchmark | thinking | Source | Trend |
|---|---|---|---|
| Epoch Capabilities Indexnot in index | 145.8#80 | Epoch AI Benchmarking Hub | |
| AA Intelligence Index v4.3.2 | 22.0#205 | Artificial Analysis |
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 |
|---|---|---|---|---|---|---|
| NovitaAI | 41 tok/s | 3.03 s | $0.600 | $2.50 | 262k | bf16 |
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 $0.150 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0010 | $0.0009 | 20.9 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0087 | $0.0046 | 23.4 s |
| Code edit | 6,000 / 1,500 | $0.0074 | $0.0053 | 30.9 s |
| Agentic coding session | 60,000 / 4,000 | $0.046 | $0.026 | 51.6 s |
| Structured extraction | 2,000 / 200 | $0.0017 | $0.0010 | 20.1 s |
More from Moonshot AI
See also
Questions people ask
- What are Kimi K2 Thinking's benchmark scores?
- As of 25 Oct 2026, Kimi K2 Thinking (thinking effort) scores 52.4 ±3.0 on the BenchLeader Index, strongest in long context (61) and weakest in coding (49). That rests on 22 independent benchmark results, and every score on this page links to its publisher.
- How does Kimi K2 Thinking rank?
- Kimi K2 Thinking (thinking effort) is #316 of 785 ranked configurations on 25 Oct 2026. 284 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 Thinking cost per 1M tokens?
- As of 25 Oct 2026, Kimi K2 Thinking costs $0.600 per 1M input tokens and $2.50 per 1M output tokens, $1.07 blended (three input tokens per output token).
- When was Kimi K2 Thinking released?
- Moonshot AI released Kimi K2 Thinking on 6 Nov 2025.
Data as of 25 Oct 2026. Compare with another model.
Cite as: BenchLeader, “Kimi K2 Thinking: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/kimi-k2-thinking.