Kimi K2 Thinking
Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning. Built on the trillion-parameter Mixture-of-Experts (MoE) architecture introduced in Kimi K2, it activates 32 billion parameters per forward pass and supports 256 k-token context windows. The model is optimized for persistent step-by-step thought, dynamic tool invocation, and complex reasoning workflows that span hundreds of turns. It interleaves step-by-step reasoning with tool use, enabling autonomous research, coding, and writing that can persist for hundreds of sequential actions without drift.
It sets new open-source benchmarks on HLE, BrowseComp, SWE-Multilingual, and LiveCodeBench, while maintaining stable multi-agent behavior through 200–300 tool calls. Built on a large-scale MoE architecture with MuonClip optimization, it combines strong reasoning depth with high inference efficiency for demanding agentic and analytical tasks.
- Organization
- Moonshot AI
- Family
- kimi-thinking
- Providers
- 16
- Context
- 262,144
- Output limit
- 262,144
- Knowledge
- 2024-08
- Release
- 2025-11-06
- Updated
- —
- Weights
- Open
- Input
- text
- Output
- text
- Capabilities
- —
Providers
Compare every cataloged offer for this model, including its provider, route, token limits, price, and capabilities.
| Provider | Model ID | Context | Output | Input / output · 1M | Reasoning | Tools | Structured | Details |
|---|---|---|---|---|---|---|---|---|
| 302.AI | kimi-k2-thinking | 262,144 | 262,144 | — / — | — | — | — | View |
| AI/ML API | moonshotai/kimi-k2-thinking | 262,144 | 98,304 | — / — | — | — | — | View |
| Amazon Bedrock | moonshot.kimi-k2-thinking | 262,144 | 16,000 | — / — | — | — | — | View |
| Charm Hyper | kimi-k2-thinking | 262,144 | 26,214 | — / — | — | — | — | View |
| Eden AI | amazon/moonshot.kimi-k2-thinking | 256,000 | 262,144 | — / — | — | — | — | View |
| Helicone | kimi-k2-thinking | 256,000 | 262,144 | — / — | — | — | — | View |
| Hugging Face | moonshotai/Kimi-K2-Thinking | 262,144 | 262,144 | — / — | — | — | — | View |
| IO.NET | moonshotai/Kimi-K2-Thinking | 32,768 | 4,096 | — / — | — | — | — | View |
| Kilo Gateway | moonshotai/kimi-k2-thinking | 262,144 | 98,304 | — / — | — | — | — | View |
| Meganova | moonshotai/Kimi-K2-Thinking | 262,144 | 262,144 | — / — | — | — | — | View |
| NanoGPT | moonshotai/kimi-k2-thinking | 262,144 | 98,304 | — / — | — | — | — | View |
| Novita AI | moonshotai/kimi-k2-thinking | 262,144 | 98,304 | — / — | — | — | — | View |
| OpenCode | kimi-k2-thinking | 262,144 | 262,144 | — / — | — | — | — | View |
| Googlevia OpenRouter | moonshotai/kimi-k2-thinking unknown | 262,144 | 235,929 | $0.60 / $2.50 | Yes | Yes | Yes | View |
| Novitavia OpenRouter | moonshotai/kimi-k2-thinking bf16 | 262,144 | 98,304 | $0.60 / $2.50 | Yes | Yes | Yes | View |
| Poe | novita/kimi-k2-thinking | 256,000 | 0 | — / — | — | — | — | View |
| Qiniu | moonshotai/kimi-k2-thinking | 256,000 | 100,000 | — / — | — | — | — | View |
Prices and limits apply to each offer. A dash means the value is not provided. OpenRouter offers show the hosting provider and routing channel separately.
Pricing
Compare token rates and cache charges across available routes and hosting configurations.
Lowest listed input and output prices across the offers above; they may belong to different providers. These are base rates, before conditional pricing, discounts or additional fees.
moonshotai/kimi-k2-thinkingObserved - Input
- $0.60
- Output
- $2.50
- Cache read
- $0.15
Hosting prices through OpenRouter All endpoints and cache rates
| Endpoint | Input / 1M | Output / 1M | Cache read / 1M | Cache write / 1M |
|---|---|---|---|---|
| Googlegoogle-vertex | $0.60 | $2.50 | — | — |
| Novitanovita/bf16 | $0.60 | $2.50 | $0.15 | — |
Each row is a hosting configuration. Open an offer in Providers for its full pricing conditions and additional fees. Endpoint observations may differ from the routing catalog's base price.
Performance
Review reported latency and output-throughput percentiles from the latest observed rolling window.
Time to first token · milliseconds
All latency percentiles 2 endpoints
| Endpoint | P50 | P75 | P90 | P99 |
|---|---|---|---|---|
| Googlegoogle-vertex | 755 ms | 908 ms | 1,222.4 ms | 2,012.28 ms |
| Novitanovita/bf16 | 1,057 ms | 2,065 ms | 3,625 ms | 21,777.56 ms |
Output throughput · tokens per second
All throughput percentiles 2 endpoints
| Endpoint | P50 | P75 | P90 | P99 |
|---|---|---|---|---|
| Googlegoogle-vertex | 161 t/s | 195 t/s | 223.6 t/s | 281.78 t/s |
| Novitanovita/bf16 | 55 t/s | 56 t/s | 57 t/s | 198.51 t/s |
Percentiles describe the observed request distribution. These are snapshots of a rolling window, not a historical time series.
Uptime
Compare successful request rates across the latest 5-minute, 30-minute, and 24-hour observations.
Reported by OpenRouter at the time of collection; rate-limited requests are excluded. A dash means no measurement was reported. This is not a live availability check.
Benchmarks
Review attributed evaluation results on each source's original scale.
moonshotai/kimi-k2-thinkingObserved | Design Arena | Category | Elo | Win rate | Rank |
|---|---|---|---|---|
| models | website | 1,118 | 48.8% | 97 |
Sources: Artificial Analysis and Design Arena, via OpenRouter. Scores retain their original scales and are not OpenAI Suite ratings.
Detailed evaluations
| Evaluation | Configuration | Source | Results | Details |
|---|---|---|---|---|
| tau bench verified airline | MoonshotAI: Kimi K2 Thinkingmoonshotai/kimi-k2-thinking-20251106 | OpenRouter | 72% | Details |
| gpqa diamond | MoonshotAI: Kimi K2 Thinkingmoonshotai/kimi-k2-thinking-20251106 | OpenRouter | 88.16% | Details |
| Composite indices | Kimi K2 Thinkingmoonshotai/kimi-k2-thinking-20251106 | Artificial Analysis | Intelligence 22 · Coding — · Agentic — | Details |
| models · website | Kimi K2 Thinkingmoonshotai/kimi-k2-thinking-20251106 | Design Arena | 1,113 Elo | Details |
Sources: Artificial Analysis, Design Arena and OpenRouter. Evaluation configurations and scales differ; results are not interchangeable. Updated Oct 24, 2026.
Technical details & API
Explore specifications, supported parameters, reasoning controls, and API identifiers for each OpenRouter route.
moonshotai/kimi-k2-thinkingObserved - Context
- 262,144
- Maximum output
- 98,304
- Tokenizer
- Other
- Input
- text
- Output
- text
- Moderation
- No
- Added to OpenRouter
- 2025-11-06
https://openrouter.ai/api/v1moonshotai/kimi-k2-thinkingmoonshotai/kimi-k2-thinking-20251106moonshotai/Kimi-K2-ThinkingReasoning
- Required
- Yes
- Enabled by default
- Yes
- Default effort
- —
- Supported efforts
- —
Supported parameters
frequency_penaltyinclude_reasoninglogprobsmax_tokenspresence_penaltyreasoningrepetition_penaltyresponse_formatseedstopstructured_outputstemperaturetool_choicetoolstop_ktop_logprobstop_p
Frequently asked questions
Answers based on the model metadata and serving offers currently in the catalog.
How much does Kimi K2 Thinking cost?
Listed rates start at $0.60 per million input tokens and $2.50 per million output tokens. Prices vary by provider and configuration; see Pricing for details.
Which providers offer Kimi K2 Thinking?
There are 17 cataloged offers across 16 providers. Compare model IDs, prices and limits in Providers.
What is the context window?
The cataloged model context is 262,144 tokens. Each serving endpoint may apply a different limit.
Does it support tools and structured output?
Tool calling: Not reported. Structured output: Not reported. Support can vary by endpoint.
Resources
Catalog history · 3 recorded revisions
Metadata observations since this model was first cataloged. These are not model release versions.
- ·
e0260d33d445 - ·
08b81e53d7d4 - ·
b53101bc7294