Kimi · Model profile
Kimi K2 Thinking
Intelligence, Performance & Price Analysis
Source: Canonical slug: kimi-k2-thinking · Canonical model registry at build time
Routing verdict
Frontier research / complex reasoning
Reserve this model for the hardest requests; simpler or repetitive work should stay on cheaper routes.
Kimi K2 Thinking summary metrics
- Intelligence
- 33 scoreKimi K2 Thinking scores 33 (estimated) on the Artificial Analysis Intelligence Index, placing it above average among other open weight models of similar size (median: 25).
- Speed
- 160.2 output tokens/secKimi K2 Thinking generates output at 160.2 tokens per second (based on the median across providers serving the model), which is well above average compared to other open weight models of similar size (median: 61.7 t/s).
- Latency
- 1.26s TTFTKimi K2 Thinking has a time to first token (TTFT) of 1.26s (based on the median across providers serving the model), which is very competitive compared to other open weight models of similar size (median: 2.01s).
- Input Price
- $0.60 / 1M tokensKimi K2 Thinking costs $0.60 per 1M input tokens (somewhat higher than average, median: $0.59) and $2.50 per 1M output tokens (somewhat higher than average, median: $2.20), based on the median across providers serving the model.
- Output Price
- $2.50 / 1M tokensKimi K2 Thinking costs $0.60 per 1M input tokens (somewhat higher than average, median: $0.59) and $2.50 per 1M output tokens (somewhat higher than average, median: $2.20), based on the median across providers serving the model.
Assessment
Routing verdict and task fit
| Workload | Rating | Notes |
|---|---|---|
| Complex reasoning & agentic workflows | Optimal | Intelligence score 33 places this model among top performers. Suitable for multi-step analysis and agentic loops. |
| High-volume chat & customer-facing | Optimal | Output speed 160.2 tokens/sec and capable intelligence make this suitable for real-time chat at scale. |
| Latency-sensitive applications | Optimal | TTFT 1.26s — adequate latency for most interactive use cases. |
| Cost-sensitive pipelines | Viable | Output pricing at $2.50 is premium. Route high-volume simple tasks to cheaper alternatives. |
Evidence
Evidence charts
AA-Omniscience Index
AA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct. · Evaluation results measured independently by Artificial Analysis
- Claude Fable 5 (with fallback)40.1
- Gemini 3.5 Flash22.7
- GPT-5.5 (xhigh)20.1
- Grok 4.3 (high)18.3
- Kimi K2.66.4
- Muse Spark4.1
- GLM-5.2 (max)4
- MiMo-V2.5-Pro3.6
- GLM-5.11.9
- MiniMax-M31.4
- MiniMax-M2.70.683
- Nemotron 3 Ultra-0.833
- Command A+-4
- DeepSeek V4 Pro (max)-10
- Kimi K2.7 Code-10.7
- Kimi K2 Thinking (this model)-20.5
- DeepSeek V4 Flash (max)-22.9
- Nex-N2-Pro-27.6
- Qwen3.5 397B A17B-29.8
- DeepSeek R1 (Jan)-31.3
| Model | AA-Omniscience Index |
|---|---|
| Claude Fable 5 (with fallback) | 40.1 |
| Gemini 3.5 Flash | 22.7 |
| GPT-5.5 (xhigh) | 20.1 |
| Grok 4.3 (high) | 18.3 |
| Kimi K2.6 | 6.4 |
| Muse Spark | 4.1 |
| GLM-5.2 (max) | 4 |
| MiMo-V2.5-Pro | 3.6 |
| GLM-5.1 | 1.9 |
| MiniMax-M3 | 1.4 |
| MiniMax-M2.7 | 0.683 |
| Nemotron 3 Ultra | -0.833 |
| Command A+ | -4 |
| DeepSeek V4 Pro (max) | -10 |
| Kimi K2.7 Code | -10.7 |
| Kimi K2 Thinking (this model) | -20.5 |
| DeepSeek V4 Flash (max) | -22.9 |
| Nex-N2-Pro | -27.6 |
| Qwen3.5 397B A17B | -29.8 |
| DeepSeek R1 (Jan) | -31.3 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Chart interpretation
- Each bar is one model measured the same way, so this model can be read against the others. The chart lists its source and retrieval date.
Artificial Analysis Intelligence Index by Open Weights / Proprietary
Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR · Evaluation results measured independently by Artificial Analysis
- Claude Fable 5 (with fallback)59.9
- GPT-5.5 (xhigh)54.8
- GLM-5.2 (max)51.1
- Gemini 3.5 Flash50.2
- MiniMax-M344.4
- DeepSeek V4 Pro (max)44.3
- Kimi K2.644.2
- Muse Spark43.1
- MiMo-V2.5-Pro42.2
- Kimi K2.7 Code41.9
- Nex-N2-Pro41
- DeepSeek V4 Flash (max)40.3
- GLM-5.140.2
- MiniMax-M2.738.1
- Nemotron 3 Ultra37.8
- Grok 4.3 (high)37.6
- GLM-4.733.7
- Qwen3.5 397B A17B33.7
- Kimi K2 Thinking (this model)32.7
- Ring-2.6-1T30.6
| Model | Artificial Analysis Intelligence Index by Open Weights / Proprietary |
|---|---|
| Claude Fable 5 (with fallback) | 59.9 |
| GPT-5.5 (xhigh) | 54.8 |
| GLM-5.2 (max) | 51.1 |
| Gemini 3.5 Flash | 50.2 |
| MiniMax-M3 | 44.4 |
| DeepSeek V4 Pro (max) | 44.3 |
| Kimi K2.6 | 44.2 |
| Muse Spark | 43.1 |
| MiMo-V2.5-Pro | 42.2 |
| Kimi K2.7 Code | 41.9 |
| Nex-N2-Pro | 41 |
| DeepSeek V4 Flash (max) | 40.3 |
| GLM-5.1 | 40.2 |
| MiniMax-M2.7 | 38.1 |
| Nemotron 3 Ultra | 37.8 |
| Grok 4.3 (high) | 37.6 |
| GLM-4.7 | 33.7 |
| Qwen3.5 397B A17B | 33.7 |
| Kimi K2 Thinking (this model) | 32.7 |
| Ring-2.6-1T | 30.6 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Benchmark interpretation
- Higher values indicate stronger overall capability. Bars show how models compare on composite intelligence metrics.
Artificial Analysis Intelligence Index
Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR · Evaluation results measured independently by Artificial Analysis
- Claude Fable 5 (with fallback)59.9
- GPT-5.5 (xhigh)54.8
- GLM-5.2 (max)51.1
- Gemini 3.5 Flash50.2
- MiniMax-M344.4
- DeepSeek V4 Pro (max)44.3
- Kimi K2.644.2
- Muse Spark43.1
- MiMo-V2.5-Pro42.2
- Kimi K2.7 Code41.9
- Nex-N2-Pro41
- DeepSeek V4 Flash (max)40.3
- GLM-5.140.2
- MiniMax-M2.738.1
- Nemotron 3 Ultra37.8
- Grok 4.3 (high)37.6
- GLM-4.733.7
- Qwen3.5 397B A17B33.7
- Kimi K2 Thinking (this model)32.7
- Ring-2.6-1T30.6
| Model | Artificial Analysis Intelligence Index |
|---|---|
| Claude Fable 5 (with fallback) | 59.9 |
| GPT-5.5 (xhigh) | 54.8 |
| GLM-5.2 (max) | 51.1 |
| Gemini 3.5 Flash | 50.2 |
| MiniMax-M3 | 44.4 |
| DeepSeek V4 Pro (max) | 44.3 |
| Kimi K2.6 | 44.2 |
| Muse Spark | 43.1 |
| MiMo-V2.5-Pro | 42.2 |
| Kimi K2.7 Code | 41.9 |
| Nex-N2-Pro | 41 |
| DeepSeek V4 Flash (max) | 40.3 |
| GLM-5.1 | 40.2 |
| MiniMax-M2.7 | 38.1 |
| Nemotron 3 Ultra | 37.8 |
| Grok 4.3 (high) | 37.6 |
| GLM-4.7 | 33.7 |
| Qwen3.5 397B A17B | 33.7 |
| Kimi K2 Thinking (this model) | 32.7 |
| Ring-2.6-1T | 30.6 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Benchmark interpretation
- Higher values indicate stronger overall capability. Bars show how models compare on composite intelligence metrics.
Artificial Analysis Openness Index: Score
Openness Index assesses model openness on a 0 to 100 normalized scale (higher is more open) · Evaluation results measured independently by Artificial Analysis
- Nemotron 3 Ultra83.3
- MiMo-V2-Flash52.8
- DeepSeek V4 Pro (max)50
- DeepSeek V4 Flash (max)50
- GLM-5.2 (max)44.4
- GLM-5.144.4
- GLM-4.744.4
- GLM-4.644.4
- Qwen3 235B A22B 250744.4
- gpt-oss-120b (high)38.9
- MiMo-V2.5-Pro38.9
- Nex-N2-Pro38.9
- Qwen3.5 397B A17B38.9
- Ring-2.6-1T38.9
- Step 3.7 Flash38.9
- Command A+38.9
- Kimi K2.633.3
- MiniMax-M333.3
- Kimi K2 Thinking (this model)27.8
- Kimi K2.7 Code27.8
| Model | Artificial Analysis Openness Index: Score |
|---|---|
| Nemotron 3 Ultra | 83.3 |
| MiMo-V2-Flash | 52.8 |
| DeepSeek V4 Pro (max) | 50 |
| DeepSeek V4 Flash (max) | 50 |
| GLM-5.2 (max) | 44.4 |
| GLM-5.1 | 44.4 |
| GLM-4.7 | 44.4 |
| GLM-4.6 | 44.4 |
| Qwen3 235B A22B 2507 | 44.4 |
| gpt-oss-120b (high) | 38.9 |
| MiMo-V2.5-Pro | 38.9 |
| Nex-N2-Pro | 38.9 |
| Qwen3.5 397B A17B | 38.9 |
| Ring-2.6-1T | 38.9 |
| Step 3.7 Flash | 38.9 |
| Command A+ | 38.9 |
| Kimi K2.6 | 33.3 |
| MiniMax-M3 | 33.3 |
| Kimi K2 Thinking (this model) | 27.8 |
| Kimi K2.7 Code | 27.8 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Chart interpretation
- Each bar is one model measured the same way, so this model can be read against the others. The chart lists its source and retrieval date.
Intelligence
Artificial Analysis Intelligence Index · Higher is better · Evaluation results measured independently by Artificial Analysis
- Claude Fable 5 (with fallback)59.9
- GPT-5.5 (xhigh)54.8
- GLM-5.2 (max)51.1
- Gemini 3.5 Flash50.2
- MiniMax-M344.4
- DeepSeek V4 Pro (max)44.3
- Kimi K2.644.2
- Muse Spark43.1
- Nemotron 3 Ultra37.8
- Grok 4.3 (high)37.6
- Kimi K2 Thinking (this model)32.7
- gpt-oss-120b (high)23.8
| Model | Intelligence |
|---|---|
| Claude Fable 5 (with fallback) | 59.9 |
| GPT-5.5 (xhigh) | 54.8 |
| GLM-5.2 (max) | 51.1 |
| Gemini 3.5 Flash | 50.2 |
| MiniMax-M3 | 44.4 |
| DeepSeek V4 Pro (max) | 44.3 |
| Kimi K2.6 | 44.2 |
| Muse Spark | 43.1 |
| Nemotron 3 Ultra | 37.8 |
| Grok 4.3 (high) | 37.6 |
| Kimi K2 Thinking (this model) | 32.7 |
| gpt-oss-120b (high) | 23.8 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Benchmark interpretation
- Higher values indicate stronger overall capability. Bars show how models compare on composite intelligence metrics.
Output Speed
Output tokens per second · Higher is better · Evaluation results measured independently by Artificial Analysis
- Step 3.7 Flash410 t/s
- gpt-oss-120b (high)314 t/s
- Nemotron 3 Ultra250 t/s
- GLM-5.2 (max)218 t/s
- Command A+196 t/s
- Gemini 3.5 Flash192 t/s
- Grok 4.3 (high)164 t/s
- Kimi K2 Thinking (this model)160 t/s
- Ring-2.6-1T136 t/s
- GLM-4.7126 t/s
- DeepSeek V4 Flash (max)112 t/s
- MiniMax-M399 t/s
- Nex-N2-Pro92 t/s
- GPT-5.5 (xhigh)88 t/s
- GLM-5.180 t/s
- Kimi K2.676 t/s
- DeepSeek V4 Pro (max)72 t/s
- Claude Fable 5 (with fallback)71 t/s
- Qwen3 235B A22B 250761 t/s
- MiMo-V2.5-Pro53 t/s
| Model | Output Speed |
|---|---|
| Step 3.7 Flash | 410 t/s |
| gpt-oss-120b (high) | 314 t/s |
| Nemotron 3 Ultra | 250 t/s |
| GLM-5.2 (max) | 218 t/s |
| Command A+ | 196 t/s |
| Gemini 3.5 Flash | 192 t/s |
| Grok 4.3 (high) | 164 t/s |
| Kimi K2 Thinking (this model) | 160 t/s |
| Ring-2.6-1T | 136 t/s |
| GLM-4.7 | 126 t/s |
| DeepSeek V4 Flash (max) | 112 t/s |
| MiniMax-M3 | 99 t/s |
| Nex-N2-Pro | 92 t/s |
| GPT-5.5 (xhigh) | 88 t/s |
| GLM-5.1 | 80 t/s |
| Kimi K2.6 | 76 t/s |
| DeepSeek V4 Pro (max) | 72 t/s |
| Claude Fable 5 (with fallback) | 71 t/s |
| Qwen3 235B A22B 2507 | 61 t/s |
| MiMo-V2.5-Pro | 53 t/s |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Output speed context
- Speed should be read together with latency and cost; fast output does not always mean better user-perceived performance.
Speed
Output tokens per second · Higher is better · Evaluation results measured independently by Artificial Analysis
- gpt-oss-120b (high)314 t/s
- Nemotron 3 Ultra250 t/s
- GLM-5.2 (max)218 t/s
- Gemini 3.5 Flash192 t/s
- Grok 4.3 (high)164 t/s
- Kimi K2 Thinking (this model)160 t/s
- MiniMax-M399 t/s
- GPT-5.5 (xhigh)88 t/s
- Kimi K2.676 t/s
- DeepSeek V4 Pro (max)72 t/s
- Claude Fable 5 (with fallback)71 t/s
| Model | Speed |
|---|---|
| gpt-oss-120b (high) | 314 t/s |
| Nemotron 3 Ultra | 250 t/s |
| GLM-5.2 (max) | 218 t/s |
| Gemini 3.5 Flash | 192 t/s |
| Grok 4.3 (high) | 164 t/s |
| Kimi K2 Thinking (this model) | 160 t/s |
| MiniMax-M3 | 99 t/s |
| GPT-5.5 (xhigh) | 88 t/s |
| Kimi K2.6 | 76 t/s |
| DeepSeek V4 Pro (max) | 72 t/s |
| Claude Fable 5 (with fallback) | 71 t/s |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Output speed context
- Speed should be read together with latency and cost; fast output does not always mean better user-perceived performance.
End-to-End Response Time
Seconds to output 500 tokens, including reasoning model 'thinking' time · Lower is better · Evaluation results measured independently by Artificial Analysis
- Step 3.7 Flash1.22 s
- gpt-oss-120b (high)1.59 s
- Nemotron 3 Ultra2 s
- GLM-5.2 (max)2.3 s
- Command A+2.55 s
- Kimi K2 Thinking (this model)3.12 s
- GLM-4.73.97 s
- Ring-2.6-1T3.67 s
- Grok 4.3 (high)3.05 s
- MiniMax-M35.05 s
- Nex-N2-Pro5.42 s
- Gemini 3.5 Flash2.6 s
- Qwen3 235B A22B 25078.14 s
- MiMo-V2.5-Pro9.52 s
- GLM-4.610.18 s
- GLM-5.16.28 s
- DeepSeek V4 Flash (max)4.46 s
- Kimi K2.7 Code9.72 s
- MiniMax-M2.79.79 s
- Kimi K2.66.56 s
| Model | End-to-End Response Time |
|---|---|
| Step 3.7 Flash | 1.22 s |
| gpt-oss-120b (high) | 1.59 s |
| Nemotron 3 Ultra | 2 s |
| GLM-5.2 (max) | 2.3 s |
| Command A+ | 2.55 s |
| Kimi K2 Thinking (this model) | 3.12 s |
| GLM-4.7 | 3.97 s |
| Ring-2.6-1T | 3.67 s |
| Grok 4.3 (high) | 3.05 s |
| MiniMax-M3 | 5.05 s |
| Nex-N2-Pro | 5.42 s |
| Gemini 3.5 Flash | 2.6 s |
| Qwen3 235B A22B 2507 | 8.14 s |
| MiMo-V2.5-Pro | 9.52 s |
| GLM-4.6 | 10.18 s |
| GLM-5.1 | 6.28 s |
| DeepSeek V4 Flash (max) | 4.46 s |
| Kimi K2.7 Code | 9.72 s |
| MiniMax-M2.7 | 9.79 s |
| Kimi K2.6 | 6.56 s |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Chart interpretation
- Each bar is one model measured the same way, so this model can be read against the others. The chart lists its source and retrieval date.
Latency: Time To First Answer Token
Seconds to first answer token received · Accounts for reasoning model 'thinking' time · Evaluation results measured independently by Artificial Analysis
- Step 3.7 Flash4.88 s
- gpt-oss-120b (high)6.37 s
- Nemotron 3 Ultra9.09 s
- Command A+10.18 s
- GLM-5.2 (max)9.19 s
- Kimi K2 Thinking (this model)12.48 s
- GLM-4.715.87 s
- Ring-2.6-1T14.68 s
- Grok 4.3 (high)0 s
- MiniMax-M320.19 s
- Nex-N2-Pro21.66 s
- Gemini 3.5 Flash0 s
- Qwen3 235B A22B 250732.55 s
- MiMo-V2.5-Pro38.08 s
- GLM-4.640.7 s
- Kimi K2.7 Code43.34 s
- GLM-5.147.62 s
- MiniMax-M2.748.19 s
- DeepSeek V4 Flash (max)50.09 s
- Kimi K2.658.43 s
| Model | Latency: Time To First Answer Token |
|---|---|
| Step 3.7 Flash | 4.88 s |
| gpt-oss-120b (high) | 6.37 s |
| Nemotron 3 Ultra | 9.09 s |
| Command A+ | 10.18 s |
| GLM-5.2 (max) | 9.19 s |
| Kimi K2 Thinking (this model) | 12.48 s |
| GLM-4.7 | 15.87 s |
| Ring-2.6-1T | 14.68 s |
| Grok 4.3 (high) | 0 s |
| MiniMax-M3 | 20.19 s |
| Nex-N2-Pro | 21.66 s |
| Gemini 3.5 Flash | 0 s |
| Qwen3 235B A22B 2507 | 32.55 s |
| MiMo-V2.5-Pro | 38.08 s |
| GLM-4.6 | 40.7 s |
| Kimi K2.7 Code | 43.34 s |
| GLM-5.1 | 47.62 s |
| MiniMax-M2.7 | 48.19 s |
| DeepSeek V4 Flash (max) | 50.09 s |
| Kimi K2.6 | 58.43 s |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Latency considerations
- Latency includes thinking / reasoning time when applicable. Lower is better for interactive use cases.
Context Window
Context window: tokens limit · Higher is better · Evaluation results measured independently by Artificial Analysis
- Gemini 3.5 Flash1M
- Claude Fable 5 (with fallback)1M
- GLM-5.2 (max)1M
- DeepSeek V4 Pro (max)1M
- Grok 4.3 (high)1M
- MiniMax-M31M
- MiMo-V2.5-Pro1M
- DeepSeek V4 Flash (max)1M
- GPT-5.5 (xhigh)922k
- Nemotron 3 Ultra262.1k
- Muse Spark262.1k
- Qwen3.5 397B A17B262.1k
- Ring-2.6-1T262.1k
- Step 3.7 Flash262.1k
- Nex-N2-Pro262k
- Kimi K2 Thinking (this model)256k
- Kimi K2.6256k
- Kimi K2.7 Code256k
- MiMo-V2-Flash256k
- Qwen3 235B A22B 2507256k
| Model | Context Window |
|---|---|
| Gemini 3.5 Flash | 1M |
| Claude Fable 5 (with fallback) | 1M |
| GLM-5.2 (max) | 1M |
| DeepSeek V4 Pro (max) | 1M |
| Grok 4.3 (high) | 1M |
| MiniMax-M3 | 1M |
| MiMo-V2.5-Pro | 1M |
| DeepSeek V4 Flash (max) | 1M |
| GPT-5.5 (xhigh) | 922k |
| Nemotron 3 Ultra | 262.1k |
| Muse Spark | 262.1k |
| Qwen3.5 397B A17B | 262.1k |
| Ring-2.6-1T | 262.1k |
| Step 3.7 Flash | 262.1k |
| Nex-N2-Pro | 262k |
| Kimi K2 Thinking (this model) | 256k |
| Kimi K2.6 | 256k |
| Kimi K2.7 Code | 256k |
| MiMo-V2-Flash | 256k |
| Qwen3 235B A22B 2507 | 256k |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Context window sizing
- Larger context windows enable complex multi-turn tasks but may increase latency and cost for long inputs.
Cost per Intelligence Index Task
Weighted average cost (USD) per Artificial Analysis Intelligence Index task, segmented by token type. Lower is better · Evaluation results measured independently by Artificial Analysis
- MiMo-V2.5-Pro0.00466
- DeepSeek V4 Pro (max)0.0063
- gpt-oss-120b (high)0.00454
- MiniMax-M30.0139
- Grok 4.3 (high)0.0101
- Kimi K2.7 Code0.0203
- GLM-5.10.0426
- Nemotron 3 Ultra0.0209
- DeepSeek R1 (Jan)0.0153
- GLM-4.60.0144
- GLM-4.70.0157
- Qwen3.5 397B A17B0.022
- Ring-2.6-1T0.0239
- Kimi K2.60.0216
- MiMo-V2-Flash0.00812
- GLM-5.2 (max)0.0268
- Gemini 3.5 Flash0.0884
- GPT-5.5 (xhigh)0.107
- Claude Fable 5 (with fallback)0.367
| Model | Cost per Intelligence Index Task |
|---|---|
| MiMo-V2.5-Pro | 0.00466 |
| DeepSeek V4 Pro (max) | 0.0063 |
| gpt-oss-120b (high) | 0.00454 |
| MiniMax-M3 | 0.0139 |
| Grok 4.3 (high) | 0.0101 |
| Kimi K2.7 Code | 0.0203 |
| GLM-5.1 | 0.0426 |
| Nemotron 3 Ultra | 0.0209 |
| DeepSeek R1 (Jan) | 0.0153 |
| GLM-4.6 | 0.0144 |
| GLM-4.7 | 0.0157 |
| Qwen3.5 397B A17B | 0.022 |
| Ring-2.6-1T | 0.0239 |
| Kimi K2.6 | 0.0216 |
| MiMo-V2-Flash | 0.00812 |
| GLM-5.2 (max) | 0.0268 |
| Gemini 3.5 Flash | 0.0884 |
| GPT-5.5 (xhigh) | 0.107 |
| Claude Fable 5 (with fallback) | 0.367 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Benchmark interpretation
- Higher values indicate stronger overall capability. Bars show how models compare on composite intelligence metrics.
- Cost pressure
- Ethen Gateway should route away from this model when the task is simple, repetitive, or likely to generate many output tokens.
Cost per Task
Weighted average cost (USD) per Intelligence Index task · Lower is better · Evaluation results measured independently by Artificial Analysis
- DeepSeek V4 Pro (max)$0.045
- gpt-oss-120b (high)$0.061
- MiniMax-M3$0.125
- Grok 4.3 (high)$0.139
- Nemotron 3 Ultra$0.244
- Kimi K2.6$0.346
- GLM-5.2 (max)$0.370
- Gemini 3.5 Flash$0.586
- GPT-5.5 (xhigh)$0.863
- Claude Fable 5 (with fallback)$2.75
| Model | Cost per Task |
|---|---|
| DeepSeek V4 Pro (max) | $0.045 |
| gpt-oss-120b (high) | $0.061 |
| MiniMax-M3 | $0.125 |
| Grok 4.3 (high) | $0.139 |
| Nemotron 3 Ultra | $0.244 |
| Kimi K2.6 | $0.346 |
| GLM-5.2 (max) | $0.370 |
| Gemini 3.5 Flash | $0.586 |
| GPT-5.5 (xhigh) | $0.863 |
| Claude Fable 5 (with fallback) | $2.75 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Cost pressure
- Ethen Gateway should route away from this model when the task is simple, repetitive, or likely to generate many output tokens.
Cost to Run Artificial Analysis Intelligence Index
Cost (USD) to run all evaluations in the Artificial Analysis Intelligence Index · Evaluation results measured independently by Artificial Analysis
- gpt-oss-120b (high)3.5
- MiMo-V2.5-Pro6.7
- DeepSeek V4 Pro (max)7.2
- MiniMax-M314.5
- Grok 4.3 (high)12.5
- GLM-4.615.6
- MiMo-V2-Flash11.4
- DeepSeek R1 (Jan)22
- Nemotron 3 Ultra33.5
- Ring-2.6-1T23.5
- Kimi K2.7 Code24.3
- Qwen3.5 397B A17B24.9
- GLM-4.716.6
- GLM-5.140.4
- GLM-5.2 (max)32.9
- Kimi K2.625
- Gemini 3.5 Flash114
- GPT-5.5 (xhigh)131.8
- Claude Fable 5 (with fallback)507.6
| Model | Cost to Run Artificial Analysis Intelligence Index |
|---|---|
| gpt-oss-120b (high) | 3.5 |
| MiMo-V2.5-Pro | 6.7 |
| DeepSeek V4 Pro (max) | 7.2 |
| MiniMax-M3 | 14.5 |
| Grok 4.3 (high) | 12.5 |
| GLM-4.6 | 15.6 |
| MiMo-V2-Flash | 11.4 |
| DeepSeek R1 (Jan) | 22 |
| Nemotron 3 Ultra | 33.5 |
| Ring-2.6-1T | 23.5 |
| Kimi K2.7 Code | 24.3 |
| Qwen3.5 397B A17B | 24.9 |
| GLM-4.7 | 16.6 |
| GLM-5.1 | 40.4 |
| GLM-5.2 (max) | 32.9 |
| Kimi K2.6 | 25 |
| Gemini 3.5 Flash | 114 |
| GPT-5.5 (xhigh) | 131.8 |
| Claude Fable 5 (with fallback) | 507.6 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Benchmark interpretation
- Higher values indicate stronger overall capability. Bars show how models compare on composite intelligence metrics.
- Cost pressure
- Ethen Gateway should route away from this model when the task is simple, repetitive, or likely to generate many output tokens.
Pricing: Cache Hit, Input, and Output
Price (USD per M Tokens) · Evaluation results measured independently by Artificial Analysis
- Input
- Output
- Cache Hit
- MiMo-V2-Flash$0.400
- DeepSeek V4 Flash (max)$0.423
- gpt-oss-120b (high)$0.900
- DeepSeek V4 Pro (max)$1.31
- MiMo-V2.5-Pro$1.31
- Step 3.7 Flash$1.39
- MiniMax-M3$1.56
- MiniMax-M2.7$1.56
- GLM-4.6$2.75
- Ring-2.6-1T$2.80
- GLM-4.7$3.33
- Nex-N2-Pro$3.25
- Kimi K2 Thinking (this model)$3.70
- Nemotron 3 Ultra$3.60
- Grok 4.3 (high)$3.95
- Qwen3.5 397B A17B$4.20
- Kimi K2.6$5.11
- Kimi K2.7 Code$5.11
- GLM-5.2 (max)$6.06
- GLM-5.1$6.06
| Model | Input | Output | Cache Hit | Total |
|---|---|---|---|---|
| MiMo-V2-Flash | $0.100 | $0.300 | Not published | $0.400 |
| DeepSeek V4 Flash (max) | $0.140 | $0.280 | $0.0028 | $0.423 |
| gpt-oss-120b (high) | $0.150 | $0.600 | $0.150 | $0.900 |
| DeepSeek V4 Pro (max) | $0.435 | $0.870 | $0.0036 | $1.31 |
| MiMo-V2.5-Pro | $0.435 | $0.870 | $0.0036 | $1.31 |
| Step 3.7 Flash | $0.200 | $1.15 | $0.040 | $1.39 |
| MiniMax-M3 | $0.300 | $1.20 | $0.060 | $1.56 |
| MiniMax-M2.7 | $0.300 | $1.20 | $0.060 | $1.56 |
| GLM-4.6 | $0.550 | $2.20 | Not published | $2.75 |
| Ring-2.6-1T | $0.300 | $2.50 | Not published | $2.80 |
| GLM-4.7 | $0.600 | $2.20 | $0.525 | $3.33 |
| Nex-N2-Pro | $0.500 | $2.50 | $0.250 | $3.25 |
| Kimi K2 Thinking (this model) | $0.600 | $2.50 | $0.600 | $3.70 |
| Nemotron 3 Ultra | $0.675 | $2.67 | $0.250 | $3.60 |
| Grok 4.3 (high) | $1.25 | $2.50 | $0.200 | $3.95 |
| Qwen3.5 397B A17B | $0.600 | $3.60 | Not published | $4.20 |
| Kimi K2.6 | $0.950 | $4.00 | $0.160 | $5.11 |
| Kimi K2.7 Code | $0.950 | $4.00 | $0.160 | $5.11 |
| GLM-5.2 (max) | $1.40 | $4.40 | $0.260 | $6.06 |
| GLM-5.1 | $1.40 | $4.40 | $0.260 | $6.06 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Chart interpretation
- Each bar is one model measured the same way, so this model can be read against the others. The chart lists its source and retrieval date.
Time per Intelligence Index Task
Weighted average decode time (minutes) per task; excludes TTFT and overhead time · Lower is better · Evaluation results measured independently by Artificial Analysis
- Nemotron 3 Ultra1.4
- Grok 4.3 (high)1.6
- Command A+1.7
- Step 3.7 Flash1.9
- gpt-oss-120b (high)1.9
- Gemini 3.5 Flash2.2
- GLM-4.72.5
- GLM-5.2 (max)2.8
- Ring-2.6-1T2.9
- GPT-5.5 (xhigh)3
- MiniMax-M33.9
- Qwen3 235B A22B 25074.4
- Claude Fable 5 (with fallback)4.8
- Nex-N2-Pro4.8
- GLM-5.14.9
- Kimi K2.7 Code5
- Qwen3.5 397B A17B5.1
- MiniMax-M2.75.5
- GLM-4.65.9
- MiMo-V2.5-Pro6.1
| Model | Time per Intelligence Index Task |
|---|---|
| Nemotron 3 Ultra | 1.4 |
| Grok 4.3 (high) | 1.6 |
| Command A+ | 1.7 |
| Step 3.7 Flash | 1.9 |
| gpt-oss-120b (high) | 1.9 |
| Gemini 3.5 Flash | 2.2 |
| GLM-4.7 | 2.5 |
| GLM-5.2 (max) | 2.8 |
| Ring-2.6-1T | 2.9 |
| GPT-5.5 (xhigh) | 3 |
| MiniMax-M3 | 3.9 |
| Qwen3 235B A22B 2507 | 4.4 |
| Claude Fable 5 (with fallback) | 4.8 |
| Nex-N2-Pro | 4.8 |
| GLM-5.1 | 4.9 |
| Kimi K2.7 Code | 5 |
| Qwen3.5 397B A17B | 5.1 |
| MiniMax-M2.7 | 5.5 |
| GLM-4.6 | 5.9 |
| MiMo-V2.5-Pro | 6.1 |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Benchmark interpretation
- Higher values indicate stronger overall capability. Bars show how models compare on composite intelligence metrics.
Model Size: Total and Active Parameters
Comparison between total model parameters and parameters active during inference · Evaluation results measured independently by Artificial Analysis
- Passive Params
- Active Params
- DeepSeek V4 Pro (max)1,600B
- MiMo-V2.5-Pro1,023B
- Kimi K2 Thinking (this model)1,000B
- Kimi K2.61,000B
- Kimi K2.7 Code1,000B
- Ring-2.6-1T1,000B
- GLM-5.2 (max)753B
- GLM-5.1744B
- DeepSeek R1 (Jan)685B
- Nemotron 3 Ultra550B
- MiniMax-M3428B
- Nex-N2-Pro397B
- Qwen3.5 397B A17B397B
- GLM-4.7357B
- GLM-4.6357B
- MiMo-V2-Flash309B
- DeepSeek V4 Flash (max)284B
- Qwen3 235B A22B 2507235B
- MiniMax-M2.7230B
- Command A+218B
| Model | Passive Params | Active Params | Total |
|---|---|---|---|
| DeepSeek V4 Pro (max) | 1,551B | 49B | 1,600B |
| MiMo-V2.5-Pro | 981B | 42B | 1,023B |
| Kimi K2 Thinking (this model) | 968B | 32B | 1,000B |
| Kimi K2.6 | 968B | 32B | 1,000B |
| Kimi K2.7 Code | 968B | 32B | 1,000B |
| Ring-2.6-1T | 937B | 63B | 1,000B |
| GLM-5.2 (max) | 713B | 40B | 753B |
| GLM-5.1 | 704B | 40B | 744B |
| DeepSeek R1 (Jan) | 648B | 37B | 685B |
| Nemotron 3 Ultra | 495B | 55B | 550B |
| MiniMax-M3 | 405B | 23B | 428B |
| Nex-N2-Pro | 380B | 17B | 397B |
| Qwen3.5 397B A17B | 380B | 17B | 397B |
| GLM-4.7 | 325B | 32B | 357B |
| GLM-4.6 | 325B | 32B | 357B |
| MiMo-V2-Flash | 294B | 15B | 309B |
| DeepSeek V4 Flash (max) | 271B | 13B | 284B |
| Qwen3 235B A22B 2507 | 213B | 22B | 235B |
| MiniMax-M2.7 | 220B | 10B | 230B |
| Command A+ | 193B | 25B | 218B |
Published benchmark dataset · Retrieved
Method and source
- Methodology
- Independent test run by Artificial Analysis on dedicated hardware.
- Architecture scale
- Larger models generally perform better but cost more to serve. Active parameter count matters for inference efficiency.
Specifications
Architecture and limits
| Specification | Value | Source |
|---|---|---|
| Model type | Open weights | faq |
| Reasoning | Yes | faq |
| Input modalities | Kimi K2 Thinking supports text input. | faq |
| Output modalities | Kimi K2 Thinking supports text output. | faq |
| Context window | 260k tokens | faq |
| Open weights / source | Yes, Kimi K2 Thinking is open weights. The model weights are publicly available and can be downloaded for self-hosting. | faq |
| Parameters | Kimi K2 Thinking has 1 trillion parameters (32 billion active). | faq |
| Active parameters | Kimi K2 Thinking is a Mixture of Experts (MoE) model with 1 trillion total parameters, but only 32 billion active parameters are used during inference. | faq |
| License | Kimi K2 Thinking is released under the Modified MIT License license. This license allows commercial use. | faq |
| API availability | Yes, Kimi K2 Thinking is available via API through 6 providers. | faq |
Method
Methodology & Provenance
This page is rendered from the normalized profile and page JSON for Kimi K2 Thinking.
Benchmark values are preserved as normalized; only layout, disclosure ordering, and typography are adjusted for readability.
Questions
Model FAQ and technical disclosures
Models
Related
- 01KimiOther Kimi models in the catalog.
- 02LeaderboardsCertified metric ranks, when published.