Moonshot AI
Chinese frontier lab behind the Kimi assistant and the Kimi K3 open-weight MoE. Kimi K3.1 Pro is the current API flagship.
Moonshot AI builds the Kimi product family and releases large open-weight models. Kimi K3 (2.8T MoE, 1M context, Kimi K3 License) remains the downloadable checkpoint, including on Databricks Unity AI Gateway (2026-08-06). Kimi K3.1 Pro (2026-10-23) is a post-trained K3 and the current API flagship (kimi-k3.1-pro and kimi-k3.1-pro-highspeed) on platform.kimi.ai. Moonshot says K3.1 Pro weights will be released by 2026-11-04.
Why Moonshot AI matters
Kimi K3 pushed open weights into the 3T-class MoE tier. K3.1 Pro is the API post-train teams call today; K3 weights remain the self-host checkpoint. Weigh capability against the custom license’s commercial clauses.
Last reviewed: 26 October 2026
Best fit for
When architects typically choose Moonshot AI.
- Open frontier multimodal models
- Long-horizon coding agents
- 1M-context knowledge work
- Self-host + API hybrid stacks
Strengths
DataAIHub qualitative scores from 1 to 5, for architects. Not official benchmark results or a public ranking.
- Open frontier scale★★★★★
- Agentic coding★★★★★
- Long context★★★★★
- Multimodal★★★★☆
- License simplicity★★☆☆☆
Quick facts
- Founded
- 2023
- Headquarters
- Beijing, China
- Open source
- Yes
- Enterprise
- No
- Flagship model
- Kimi K3
On DataAIHub
- 1Models
- 4Products
- 4Tools
- 1GitHub
- 2Research
- 8Guides
- 4Benchmarks
Company profile
- Founded
- 2023
- Headquarters
- Beijing, China
- Country
- China
- Open source
- Yes
- Cloud provider
- No
- Website
- https://www.moonshot.ai
- Confidence
- High
- Source coverage
- 9
Ecosystem
Competes with
Works with
Recommended for
- Large Language Models
Kimi K3 is a key open 3T-class reference in the 2026 LLM landscape.
- Llama Models
Compare open-weight licensing and deployability vs Llama stacks.
Often paired with
- Hugging Face
Kimi K3 weights and community serving recipes distribute via the Hub.
How Moonshot AI evolved
Key moments in chronological order.
- Model
Kimi K3.1 Pro (kimi-k3.1-pro) and Highspeed (kimi-k3.1-pro-highspeed) launch on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API. Moonshot describes K3.1 Pro as a post-trained K3 and says weights will be released by 2026-11-04. Enterprise access is off by default. Moonshot says the open-weight release removes cyber-related mid-training data and RL environments. Kimi K3 weights on Hugging Face remain the downloadable checkpoint.
- Partnership
Kimi K3 on Databricks Unity AI Gateway
Databricks hosts Kimi K3 through Foundation Model API with Unity AI Gateway governance (US hosting; AWS and GCP workspaces).
- Open source
Full checkpoint published on Hugging Face under the Kimi K3 License.
- Model
2.8T MoE multimodal agentic model ships on Kimi product and API.
- API
platform.kimi.ai
OpenAI/Anthropic-compatible API for Kimi K3 and related models.
- Founded
Moonshot AI founded
Beijing lab builds the Kimi assistant and frontier MoE models.
Products
Foundation models
Related tools
Related research
- Introducing Kimi K3.1 Pro
Original Paper
- Kimi K3 model card
Original Paper
GitHub
Related benchmarks
Related rankings
Related guides
Explore more companies
- DeepSeekChinaDeepSeek releases competitive open-weight models—including DeepSeek V4.1-Flash (live Flash, native multimodal), V4-Pro-0813 (still on deepseek-v4-pro after 2026-09-14), DeepSeek V3, and the reasoning-focused DeepSeek R1—plus APIs, chat, and DeepSeek Harness (MIT agent runtime, developer preview). DeepInfer (Apache-2.0, 2026-10-20) open-sources offline inference for DeepSeek-V4.1-Flash on NVIDIA SM100 and Huawei Ascend 950; that release does not include a KV cache, prefill/decode disaggregation, or replica load balancing. It forced the industry to revisit cost/performance assumptions for frontier-class open models, especially for reasoning and coding workloads.
- AlibabaChinaGlobal technology group whose Qwen team releases competitive multilingual and multimodal foundation models for cloud and open-weight use.
- MetaUnited StatesMeta advances open multimodal research and releases the Llama open-weight model family used widely in self-hosted AI stacks. Muse Wonder (muse-wonder-1.0, 2026-10-12) is the newer paid Meta Model API peak for longer agentic work, with Muse Code. Muse Spark 1.3 remains the lower-priced closed API. Muse Spark Open (2026-10-15) is a separate weight release: 816B total / 42B active, 1M context, multimodal, under the Muse Model License with limited commercial terms. Combined with PyTorch, Llama still lets organizations train, fine-tune, and serve models without depending on a proprietary API.
- THUDM / TsinghuaChinaTsinghua University Knowledge Engineering Group (THUDM) — research lab behind LongBench and other influential open LLM evaluation and model work.
- Mistral AIFranceMistral AI builds Mistral Large 3 (with Mistral Large 4, a 1.05T-parameter multimodal MoE with 52B active parameters: API public preview since 2026-10-06, weights published 2026-10-29 under the Mistral Ecosystem Open-Weight License), Small/Medium commercial tiers, and open-weight models, plus La Plateforme APIs and Le Chat. Mixtral remains an Apache-2.0 self-host option; it is not listed on Mistral serverless API pricing as of 2026-08-16. Regional Endpoints (EU or US) are generally available. Series D (€3B, 2026-09-08) funds sovereign open-weight infrastructure; a 2026-09-10 Cloudera partnership targets hybrid/air-gapped inference and custom models on customer data. Firefox Smart Window (beta, 2026-09-16) uses Mistral models in France and North America with zero data retention. Timestral (2026-10-15) is a separate time-series family (5M, 28M, 123M, and 209M parameters, Modified MIT), not a Large successor. Voxtral 3 (2026-10-20) is a speech-to-text and text-to-speech family plus a voice-agent SDK, available on Mistral’s API or for self-deployment, and is not a chat model.
- ReflectionUnited StatesReflection released Beam 1.0 on 2026-10-16: an Apache-2.0 text-only sparse MoE with 501B total parameters, 23B active, and a 1M-token context window, aimed at agentic coding and reasoning. Weights shipped with day-0 support in Hugging Face Transformers, vLLM, and SGLang. No first-party token price was published with the weight release.