Kimi K3
Summary
Kimi K3 is an open-weight language model from Moonshot AI, released on 16 Jul 2026. It accepts text, images and video and generates text, with a 1M-token context window and up to 1M output tokens per response. As of 18 Nov 2026, the lowest listed API price is $2 per million input tokens and $10 per million output tokens (Fireworks AI), across 13 providers we track.
Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work
Last verified 18 Nov 2026. Every figure on this page links to its source.
Specifications
| Developer | Moonshot AI |
|---|---|
| Release date | 16 Jul 2026 [models.dev] |
| Status | Generally available [models.dev] |
| Weights | Open weights [Hugging Face] |
| License | kimi-k3 [Hugging Face] |
| Parameters | 2.8T [Hugging Face] |
| Context window | 1M (1,048,576 tokens) [models.dev] |
| Max output | 1M tokens [models.dev] |
| Input | Text, images and video [models.dev] |
| Output | Text [models.dev] |
| Reasoning mode | Yes [models.dev] |
| Tool calling | Yes [models.dev] |
| Structured output | Yes [models.dev] |
| Hugging Face | moonshotai/Kimi-K3 [Hugging Face] |
Kimi K3 API pricing by provider
USD per million tokens. Sorted by input price.
| Provider | Input | Output | Cached input | Context | Source |
|---|---|---|---|---|---|
| OpenRouter moonshotai/kimi-k3 | $0.19 | $16 | $0.18 | 1M | OpenRouter, 18 Nov 2026 |
| Fireworks AI accounts/fireworks/routers/kimi-latest | $2 | $10 | $0.20 | 1M | models.dev, 18 Nov 2026 |
| Fireworks AI accounts/fireworks/routers/kimi-k3-spot | $2.10 | $10.50 | $0.21 | 1M | models.dev, 18 Nov 2026 |
| SiliconFlow moonshotai/Kimi-K3 | $2.70 | $13.50 | $0.27 | 1M | models.dev, 18 Nov 2026 |
| DeepInfra moonshotai/Kimi-K3 | $2.85 | $14.25 | $0.28 | 1M | models.dev, 18 Nov 2026 |
| Moonshot AI API kimi-k3 | $3 | $15 | $0.30 | 1M | models.dev, 18 Nov 2026 |
| Alibaba Cloud Model Studio kimi-k3 | $3 | $15 | $0.30 | 1M | models.dev, 18 Nov 2026 |
| Amazon Bedrock global.moonshotai.kimi-k3 | $3 | $15 | $0.30 | 1M | models.dev, 18 Nov 2026 |
| Fireworks AI accounts/fireworks/models/kimi-k3 | $3 | $15 | $0.30 | 1M | models.dev, 18 Nov 2026 |
| Baseten moonshotai/Kimi-K3 | $3 | $15 | $0.30 | 1M | models.dev, 18 Nov 2026 |
| Hugging Face Inference Providers moonshotai/Kimi-K3 | $3 | $15 | – | 1M | models.dev, 18 Nov 2026 |
| Nebius AI Studio moonshotai/Kimi-K3 | $3 | $15 | $0.30 | 1M | models.dev, 18 Nov 2026 |
| Novita AI moonshotai/kimi-k3 | $3 | $15 | $0.30 | 1M | models.dev, 18 Nov 2026 |
| Together AI moonshotai/Kimi-K3 | $3 | $15 | $0.30 | 1M | models.dev, 18 Nov 2026 |
| Vercel AI Gateway moonshotai/kimi-k3 | $3 | $15 | $0.30 | 1M | models.dev, 18 Nov 2026 |
| Amazon Bedrock us.moonshotai.kimi-k3 | $3.30 | $16.50 | $0.33 | 1M | models.dev, 18 Nov 2026 |
| Fireworks AI accounts/fireworks/routers/kimi-fast-latest | $4.50 | $22.50 | $0.45 | 1M | models.dev, 18 Nov 2026 |
| Fireworks AI accounts/fireworks/routers/kimi-k3-fast | $4.50 | $22.50 | $0.45 | 1M | models.dev, 18 Nov 2026 |
| Vercel AI Gateway moonshotai/kimi-k3-fast | $6 | $22.50 | $0.60 | 1M | models.dev, 18 Nov 2026 |
Running Kimi K3 locally
Estimated memory, from the parameter count and the architecture in config.json on Hugging Face. Includes about 0.8 GB of runtime overhead.
| Quantization | Weights | Total, 8K context | Total, 32K context |
|---|---|---|---|
| Q4_K_M (4-bit) | 1685 GB | 1687 GB | 1690 GB |
| Q8_0 (8-bit) | 2954 GB | 2955 GB | 2958 GB |
| FP16 / BF16 (16-bit) | 5560 GB | 5562 GB | 5564 GB |
At 4-bit with an 8K conversation, Kimi K3 needs about 1687 GB, more than any single consumer graphics card; it needs a unified-memory machine, a multi-GPU setup or a split between graphics card and system RAM. Check your own hardware, or see Best computer for local AI on our sister site AIPCs.
Frequently asked questions
What is the context window of Kimi K3?
Kimi K3 has a context window of 1,048,576 tokens (1M) and can generate up to 1,048,576 tokens in a single response, according to models.dev.
How much does the Kimi K3 API cost?
Through Moonshot AI API, Kimi K3 costs $3 per million input tokens and $15 per million output tokens. The lowest listed price is $2 input / $10 output through Fireworks AI. Prices last verified 18 Nov 2026.
Is Kimi K3 open source?
Kimi K3 is an open-weight model: its weights are published on Hugging Face as moonshotai/Kimi-K3 under the kimi-k3 license. Check the license terms before commercial use.
When was Kimi K3 released?
Moonshot AI released Kimi K3 on 16 Jul 2026, according to models.dev.
Which providers offer Kimi K3?
We track Kimi K3 on 13 providers: OpenRouter, Fireworks AI, SiliconFlow, DeepInfra, Moonshot AI API, Alibaba Cloud Model Studio, Amazon Bedrock, Baseten, Hugging Face Inference Providers, Nebius AI Studio, Novita AI, Together AI, Vercel AI Gateway.
How much VRAM does Kimi K3 need?
About 1687 GB at 4-bit (Q4_K_M) with an 8K-token conversation, and about 2955 GB at 8-bit. That is more than any single consumer graphics card holds. These are estimates from the model's published parameter count and architecture.
Similar models from other labs
Comparable on output type, weights, release date, price and context window. Prices are the lowest listed, USD per million tokens.
| Model | Lab | Released | Context | Input $/1M | Output $/1M | Weights |
|---|---|---|---|---|---|---|
| GPT-6 Sol Pro | OpenAI | 22 Sept 2026 | 1.1M | $2 | $10 | – |
| GPT-6.1 Sol Pro | OpenAI | 29 Sept 2026 | 1.1M | $2 | $10 | – |
| MiMo-V2.6-Pro-UltraSpeed | Xiaomi (MiMo) | 21 Sept 2026 | 1M | $4.35 | $8.70 | Open |
| GLM 5.3 Prime | Zhipu AI (Z.ai) | 23 Sept 2026 | 1M | $1.68 | $5.28 | – |
| Macaron V1 Venti | Mindlab | 21 Jul 2026 | 1M | $0.82 | $2.48 | Open |
| Muse Spark Open | Meta | 15 Oct 2026 | 1M | $1.25 | $4.25 | Open |
Change history
- Fireworks AI price changed: input $3 → $2, output $15 → $10 per 1M tokens
- Vercel AI Gateway price changed: input $4.50 → $6, output $22.50 → $22.50 per 1M tokens
- OpenRouter price changed: input $0.19 → $0.19, output $15 → $16 per 1M tokens
- OpenRouter price changed: input $0.24 → $0.19, output $10.80 → $15 per 1M tokens
- OpenRouter price changed: input $0.17 → $0.24, output $20 → $10.80 per 1M tokens
- OpenRouter price changed: input $0.19 → $0.17, output $20 → $20 per 1M tokens
- OpenRouter price changed: input $0.16 → $0.19, output $20 → $20 per 1M tokens
- OpenRouter price changed: input $0.24 → $0.16, output $10.80 → $20 per 1M tokens
- OpenRouter price changed: input $0.24 → $0.24, output $11.94 → $10.80 per 1M tokens
- Nebius AI Studio price changed: input $3 → $3, output $15 → $15 per 1M tokens
- Now listed on Fireworks AI
- No longer listed on nvidia
- Baseten price changed: input $3 → $3, output $15 → $15 per 1M tokens
- OpenRouter price changed: input $0.30 → $0.24, output $11.94 → $11.94 per 1M tokens
- OpenRouter price changed: input $0.45 → $0.30, output $15 → $11.94 per 1M tokens
- OpenRouter price changed: input $0.70 → $0.45, output $15 → $15 per 1M tokens
- OpenRouter price changed: input $0.73 → $0.70, output $10 → $15 per 1M tokens
- OpenRouter price changed: input $0.80 → $0.73, output $10 → $10 per 1M tokens
- OpenRouter price changed: input $0.64 → $0.80, output $13.50 → $10 per 1M tokens
- OpenRouter price changed: input $0.50 → $0.64, output $12 → $13.50 per 1M tokens
- OpenRouter price changed: input $0.62 → $0.50, output $12.30 → $12 per 1M tokens
- OpenRouter price changed: input $0.62 → $0.62, output $15 → $12.30 per 1M tokens
- OpenRouter price changed: input $0.95 → $0.62, output $14 → $15 per 1M tokens
- OpenRouter price changed: input $0.67 → $0.95, output $14 → $14 per 1M tokens
- OpenRouter price changed: input $0.72 → $0.67, output $14 → $14 per 1M tokens
Sources
- https://openrouter.ai/moonshotai/kimi-k3
- https://github.com/sst/models.dev/blob/dev/providers/moonshotai/models/kimi-k3.toml
- https://huggingface.co/moonshotai/Kimi-K3
- https://github.com/sst/models.dev/blob/dev/providers/fireworks-ai/models/accounts/fireworks/routers/kimi-latest.toml
- https://github.com/sst/models.dev/blob/dev/providers/fireworks-ai/models/accounts/fireworks/routers/kimi-k3-spot.toml
- https://github.com/sst/models.dev/blob/dev/providers/siliconflow/models/moonshotai/Kimi-K3.toml
- https://github.com/sst/models.dev/blob/dev/providers/deepinfra/models/moonshotai/Kimi-K3.toml
- https://github.com/sst/models.dev/blob/dev/providers/alibaba/models/kimi-k3.toml
- https://github.com/sst/models.dev/blob/dev/providers/amazon-bedrock/models/global.moonshotai.kimi-k3.toml
- https://github.com/sst/models.dev/blob/dev/providers/fireworks-ai/models/accounts/fireworks/models/kimi-k3.toml
- https://github.com/sst/models.dev/blob/dev/providers/baseten/models/moonshotai/Kimi-K3.toml
- https://github.com/sst/models.dev/blob/dev/providers/huggingface/models/moonshotai/Kimi-K3.toml
- https://github.com/sst/models.dev/blob/dev/providers/nebius/models/moonshotai/Kimi-K3.toml
- https://github.com/sst/models.dev/blob/dev/providers/novita-ai/models/moonshotai/kimi-k3.toml
- https://github.com/sst/models.dev/blob/dev/providers/togetherai/models/moonshotai/Kimi-K3.toml
- https://github.com/sst/models.dev/blob/dev/providers/vercel/models/moonshotai/kimi-k3.toml
- https://github.com/sst/models.dev/blob/dev/providers/amazon-bedrock/models/us.moonshotai.kimi-k3.toml
- https://github.com/sst/models.dev/blob/dev/providers/fireworks-ai/models/accounts/fireworks/routers/kimi-fast-latest.toml
- https://github.com/sst/models.dev/blob/dev/providers/fireworks-ai/models/accounts/fireworks/routers/kimi-k3-fast.toml
- https://github.com/sst/models.dev/blob/dev/providers/vercel/models/moonshotai/kimi-k3-fast.toml