token.app › Moonshot AI › Kimi K3
Kimi K3 pricing & benchmarks
Kimi K3 is available from Moonshot AI at $0.239 per million input tokens and $11.94 per million output tokens ($3.16 blended at 3:1). It accepts up to 1,048,576 tokens of context. Kimi K3 is a 2.8T parameter open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at...
Pricing
Per million tokens · OpenRouter
Input$0.239
Output$11.94
Blended (3:1)$3.16
Specification
moonshotai/kimi-k3
Context window1.0M
Max output944K
ReleasedJul 2026
LicenceOpen weights
Modalitiesin:text, in:image, in:video, out:text
Benchmarks
Independently run — not vendor self-reports. Every score links its source.
| Benchmark | Score | Config | Run | Source |
|---|---|---|---|---|
| GPQA Diamond | 93.1% ±1.5 | Max effort | 2026-07-16 | Independently run · Epoch AI |
| FrontierMath T1–3 | 72.2% ±2.7 | Max effort | 2026-07-17 | Independently run · Epoch AI |
| FrontierMath T4 | 39.0% ±7.7 | Max effort | 2026-07-17 | Independently run · Epoch AI |
| OTIS Mock AIME | 97.2% ±1.1 | Max effort | 2026-07-16 | Independently run · Epoch AI |
Scores published by Epoch AI, 'AI Benchmarking Hub', used under CC BY 4.0.
Cheaper models that score at least as well on GPQA Diamond
Strictly lower blended price AND an equal-or-higher GPQA Diamond score. One benchmark is one dimension — a model that wins here may still be weaker at your specific task.
| Model | Provider | Blended $/1M | GPQA Diamond | |
|---|---|---|---|---|
| Gemini 3.7 Flash | $1.50 (−53%) | 94.8% | Compare → | |
| Gemini 3.6 Flash | $1.50 (−53%) | 94.1% | Compare → | |
| Grok 4.6 | SpaceXAI | $3.00 (−5%) | 94.0% | Compare → |
| Grok 4.5 | SpaceXAI | $3.00 (−5%) | 93.4% | Compare → |
Where to run Kimi K3
28 options across 20 providers · 4.0× between cheapest and dearest. Blended at 3:1. ⚠ marks an option serving less context than the best available, or re-pricing above a prompt-length threshold. Bracketed labels are the provider's own service tier or region.
| Host | Blended $/1M | In / Out | Context | Weights | Uptime 24h |
|---|---|---|---|---|---|
| Sail Research (fp4/us) | $2.27 | $1.20 / $5.50 | 1.0M | fp4 | — |
| Sail Research | $2.27 | $1.20 / $5.50 | 1.0M | fp4 | — |
| InferenceNet (us) | $3.16 | $0.239 / $11.94 | 1.0M | fp4 | 100.0% |
| InferenceNet | $3.16 | $0.239 / $11.94 | 1.0M | fp4 | 100.0% |
| Relace | $3.23 | $0.300 / $12.00 | 1.0M | fp4 | 99.5% |
| Wafer (us) | $3.25 | $0.360 / $11.90 | 1.0M | mxfp4 | 100.0% |
| Decart | $3.30 | $1.65 / $8.25 | 1.0M | mxfp4 | 99.9% |
| Chutes | $3.76 | $1.85 / $9.50 | 1.0M | mxfp4 | 88.4% |
| Wafer | $4.18 | $0.238 / $16.00 | 1.0M | mxfp4 | 100.0% |
| Fireworks (spot) | $4.20 | $2.10 / $10.50 | 1.0M | — | 60.2% |
| Makora | $4.24 | $0.650 / $15.00 | 1.0M | fp4 | 99.6% |
| Morph | $4.40 | $0.894 / $14.90 | 1.0M | fp8 | 98.6% |
| AkashML | $4.40 | $1.20 / $14.00 | 1.0M | fp4 | 99.5% |
| InferenceNet (fast) | $4.83 | $2.04 / $13.20 | 1.0M | fp4 | 99.4% |
| Phala | $5.10 | $2.55 / $12.75 | 1.0M | — | 100.0% |
| DigitalOcean | $5.15 | $2.55 / $12.95 | 1.0M | — | 100.0% |
| InferenceNet (us/fast) | $5.31 | $2.24 / $14.52 | 1.0M | fp4 | 99.7% |
| DeepInfra | $5.70 | $2.85 / $14.25 | 1.0M | mxfp4 | 99.9% |
| Amazon Bedrock (us-east-2) | $6.00 | $3.00 / $15.00 | 1.0M | — | 92.3% |
| Parasail | $6.00 | $3.00 / $15.00 | 1.0M | fp4 | 95.9% |
| Modal | $6.00 | $3.00 / $15.00 | 1.0M | mxfp4 | 100.0% |
| Together | $6.00 | $3.00 / $15.00 | 1.0M | — | 100.0% |
| Fireworks | $6.00 | $3.00 / $15.00 | 1.0M | — | 98.4% |
| BaseTen | $6.00 | $3.00 / $15.00 | 1.0M | fp8 | 100.0% |
| Moonshot AI | $6.00 | $3.00 / $15.00 | 1.0M | mxfp4 | 100.0% |
| Alibaba | $6.90 | $3.45 / $17.25 | 1.0M | — | 98.7% |
| Fireworks (us) | $9.00 | $4.50 / $22.50 | 1.0M | — | 94.4% |
| Fireworks (fast) | $9.00 | $4.50 / $22.50 | 1.0M | — | 98.2% |
Host prices and uptime from OpenRouter, synced 2026-10-22. Lower-precision weights (fp4, fp8) can cost less and score worse than the same model at bf16 — the score above is not host-specific.
Subscriptions that include Kimi K3
Consumer plans whose published model list names this model
Compare Kimi K3
FAQ
How much does Kimi K3 cost?
Kimi K3 costs $0.239 per million input tokens and $11.94 per million output tokens on OpenRouter. At a 3:1 input:output mix that blends to $3.16 per million tokens.
What is Kimi K3's context window?
Kimi K3 accepts up to 1,048,576 tokens of context, and can generate up to 943,718 output tokens.
How good is Kimi K3 on benchmarks?
Kimi K3 scores 93.1% on GPQA Diamond, independently run and published by Epoch AI. 4 benchmark results are listed on this page, each with its run date and source.
Is there a cheaper model as good as Kimi K3?
Yes — 4 models in our catalogue cost less per token than Kimi K3 and score at least as high on the same benchmark. The cheapest is Gemini 3.7 Flash at $1.50 per million tokens blended.