Moonshot AIOpen SourceCodingReasoningMultimodal
Kimi K3
Moonshot’s Kimi K3 — 2.8T MoE (104B active) open-weight multimodal agentic model with 1M context, native vision, and strong long-horizon coding. Weights on Hugging Face under the Kimi K3 License.
Vision · Tool calling · Thinking · MCP · Coding
Last reviewed: 31 July 2026
Overview
Moonshot’s Kimi K3 — 2.8T MoE (104B active) open-weight multimodal agentic model with 1M context, native vision, and strong long-horizon coding. Weights on Hugging Face under the Kimi K3 License.
Capabilities
- Vision: Yes
- Audio: No
- Tool calling: Yes
- Thinking: Yes
- MCP: Yes
- Coding: Yes
- Structured output: Yes
Technical specifications
- Provider
- Moonshot AI
- License
- Kimi K3 License (custom; commercial scale conditions)
- Context window
- 1.0M
- Parameters
- MoE 2.8T total / 104B activated
- Architecture
- Stable LatentMoE (KDA + Attention Residuals)
- Release
- 2026-07
- Modalities
- Text, Image, Video
- Vision
- Yes
- Audio
- No
- Tool calling
- Yes
- Thinking
- Yes
- MCP
- Yes
- Open weights
- Yes
- API
- Yes
- Pricing (input)
- Moonshot / Kimi API (verify platform.kimi.ai)
- Pricing (output)
- Moonshot / Kimi API (verify platform.kimi.ai)
Supported modalities
Text · Image · Video
Context window
1.0M (1,048,576 tokens)
Pricing
Input: Moonshot / Kimi API (verify platform.kimi.ai)
Output: Moonshot / Kimi API (verify platform.kimi.ai)
Also self-host via open weights; large-scale MaaS may need a separate Moonshot agreement.
Availability
API: Yes
Chat UI: Yes
Open weights: Yes
API model id kimi-k3; weights at huggingface.co/moonshotai/Kimi-K3.
Use cases
- Long-horizon agentic coding
- Open multimodal research and products
- 1M-context knowledge work
- Self-hosted frontier open deployments
Strengths
- Largest open 3T-class MoE release to date
- Native multimodal + 1M context
- Competitive agentic coding vs closed frontier peers
Limitations
- Custom license — not MIT/Apache; check commercial clauses
- Very large download / serving footprint
- Ecosystem younger than Llama/Qwen stacks
Related guides
Related benchmarks
Related research
- Kimi K3 model card (Hugging Face)
Original Paper
Related GitHub
Related tools
Related rankings
Companies
Explore more models
- Qwen3AlibabaAlibaba’s Qwen3 generation — strong multilingual open models spanning chat, reasoning modes, coding, and multimodal variants.
- Llama 4MetaMeta’s Llama 4 family — open-weight multimodal models designed for research and commercial use under Meta’s community license.
- Claude FableAnthropicAnthropic’s Claude Fable 5 — the most capable widely released Claude for long-horizon agents, deep reasoning, and demanding coding workflows. Mythos 5 is the limited-access peer for Project Glasswing.
- Claude OpusAnthropicAnthropic’s Claude Opus 5 tier for complex agentic coding, enterprise work, long-context analysis, and careful instruction following. Claude Fable 5 sits above Opus for peak widely released capability.
- Claude SonnetAnthropicAnthropic’s Claude Sonnet 5 tier — best combination of speed and intelligence for most production agents and coding, at lower cost than Opus.
- Gemini 2.5 ProGoogleGoogle’s Pro-class Gemini model for advanced reasoning and native multimodal workloads across text, images, audio, and long context.