Moonshot AI models

Beijing-based AI company behind the Kimi assistant and the Kimi model family, including the open-weights Kimi K2 and K3 mixture-of-experts models.

Beijing, ChinaFounded 2023moonshot.ai

Releases
15
Open weights
14
Latest
Oct 24, 2026

October 2026

1 release
  1. LLMOpen weights

    Kimi K3.1 Pro

    Kimi K3.1 Pro is the first model in Moonshot AI's K3.1 series, a post-trained version of Kimi K3 built with substantially larger post-training compute. It improves coding, 3D creation, financial research, and other agentic work.

    Modalities
    text, image → text
    Context
    1M context
    Parameters
    MoE: 2.8T total, 104B active
    License
    Kimi K3.1 Pro License
    At launch
    API, Consumer app, Open weights
    • Released with Kimi K3.1 Pro Highspeed, which runs about 2 to 3 times faster
    • Available on Kimi.com, Kimi Work, Kimi Code, and the Kimi API as kimi-k3.1-pro
    • Weights published on Hugging Face under the Kimi K3.1 Pro License
    Source: Introducing Kimi K3.1 Pro

July 2026

1 release
  1. LLMOpen weights

    Kimi K3

    Kimi K3 is Moonshot AI's 2.8T-parameter model, which Moonshot describes as the first open 3T-class model. It is built on Kimi Delta Attention and Attention Residuals and targets long-horizon coding, knowledge work, and reasoning.

    Modalities
    text, image, video → text
    Context
    1M context
    Parameters
    MoE: 2.8T total, 104B active
    License
    Kimi K3 License
    At launch
    API, Consumer app, Open weights
    • Activates 16 of 896 experts per token
    • Natively understands text, images, and video
    • Launched on Kimi.com, Kimi Work, Kimi Code, and the Kimi API, with full weights promised by July 27, 2026
    • Weights and code under the Kimi K3 License
    Source: Kimi K3: Open Frontier Intelligence

June 2026

1 release
  1. CodingOpen weights

    Kimi K2.7 Code

    Kimi K2.7 Code is a coding-focused agentic model built on Kimi K2.6, with improvements on real-world long-horizon coding tasks. Moonshot says it uses about 30% fewer thinking tokens than K2.6.

    Modalities
    text, image → text
    Context
    250K context
    Parameters
    MoE: 1T total, 32B active
    License
    Modified MIT License
    At launch
    API, Open weights
    • Released through the Kimi Open Platform API, followed by a HighSpeed variant
    • MoonViT vision encoder with 400M parameters
    • Weights on Hugging Face under the Modified MIT License
    Source: moonshotai/Kimi-K2.7-Code (Hugging Face)

April 2026

1 release
  1. LLMOpen weights

    Kimi K2.6

    Kimi K2.6 is Moonshot AI's open-weights model focused on long-horizon coding, long-running execution, and agent swarm capabilities. It improves on Kimi K2.5 across programming languages and tasks such as front-end, DevOps, and performance optimization.

    Modalities
    text, image, video → text
    Context
    250K context
    Parameters
    MoE: 1T total, 32B active
    License
    Modified MIT License
    At launch
    API, Consumer app, Open weights
    • Available on Kimi.com, the Kimi App, the API, and Kimi Code
    • Native multimodal model with the 400M-parameter MoonViT vision encoder
    • Weights and code under the Modified MIT License
    Source: Kimi K2.6: Advancing Open-Source Coding

January 2026

1 release
  1. MultimodalOpen weights

    Kimi K2.5

    Kimi K2.5 builds on Kimi K2 with continued pre-training on about 15T mixed visual and text tokens, making it a native multimodal model. It introduced a self-directed agent swarm that can coordinate up to 100 sub-agents.

    Modalities
    text, image, video → text
    Context
    250K context
    Parameters
    MoE: 1T total, 32B active
    License
    Modified MIT License
    At launch
    API, Consumer app, Open weights
    • Agent swarm runs parallel workflows across up to 1,500 tool calls
    • Kimi.com and the Kimi App offer Instant, Thinking, Agent, and Agent Swarm (beta) modes
    • Weights and code under the Modified MIT License
    Source: Kimi K2.5: Visual Agentic Intelligence

November 2025

1 release
  1. ReasoningOpen weights

    Kimi K2 Thinking

    Kimi K2 Thinking is Moonshot AI's open-weights thinking agent model, which reasons step by step while using tools. Moonshot says it can execute 200 to 300 sequential tool calls without human intervention.

    Modalities
    text → text
    Context
    250K context
    Parameters
    MoE: 1T total, 32B active
    License
    Modified MIT License
    At launch
    API, Consumer app, Open weights
    • Scales both thinking tokens and tool-calling steps for test-time scaling
    • Live in chat mode on kimi.com and through the Kimi K2 Thinking API
    • Released with a Turbo variant on the Kimi Open Platform
    • Weights and code under the Modified MIT License
    Source: Introducing Kimi K2 Thinking

October 2025

1 release
  1. LLMOpen weights

    Kimi Linear

    Kimi Linear is a hybrid linear attention architecture built on Kimi Delta Attention, a refined version of Gated DeltaNet. Moonshot released 48B base and instruct checkpoints trained on 5.7T tokens.

    Modalities
    text → text
    Context
    1M context
    Parameters
    MoE: 48B total, 3B active
    License
    MIT
    At launch
    Open weights
    • Reduces KV cache needs by up to 75%
    • Up to 6 times higher decoding throughput at contexts up to 1M tokens
    • KDA kernel open-sourced in the flash-linear-attention library
    Source: moonshotai/Kimi-Linear-48B-A3B-Instruct (Hugging Face)

September 2025

1 release
  1. LLMOpen weights

    Kimi K2 Instruct 0905

    Kimi-K2-Instruct-0905 is an updated Kimi K2 instruct model that extends the context window from 128K to 256K tokens. Moonshot also improved its agentic coding and front-end programming abilities.

    Modalities
    text → text
    Context
    250K context
    Parameters
    MoE: 1T total, 32B active
    License
    Modified MIT License
    At launch
    API, Open weights
    • Context length doubled to 256K tokens
    • Offered on the Kimi Open Platform as kimi-k2-0905-preview
    • Weights and code under the Modified MIT License
    Source: moonshotai/Kimi-K2-Instruct-0905 (Hugging Face)

July 2025

1 release
  1. LLMOpen weights

    Kimi K2

    Kimi K2 is Moonshot AI's open-weights mixture-of-experts model optimized for agentic tasks, released as Kimi-K2-Base for fine-tuning and Kimi-K2-Instruct for chat and agent use. Moonshot trained it with the MuonClip optimizer.

    Modalities
    text → text
    Context
    125K context
    Parameters
    MoE: 1T total, 32B active
    License
    Modified MIT License
    At launch
    API, Consumer app, Open weights
    • 1T total parameters with 32B activated
    • 128K context length at launch
    • Offered on the Kimi Open Platform as kimi-k2-0711-preview
    • Weights and code under the Modified MIT License
    Source: Kimi K2: Open Agentic Intelligence

June 2025

2 releases
  1. MultimodalOpen weights

    Kimi-VL-A3B-Thinking-2506

    Kimi-VL-A3B-Thinking-2506 is an improved version of Moonshot AI's Kimi-VL-A3B-Thinking vision-language reasoning model. Moonshot recommends using it instead of the earlier Thinking checkpoint released in April 2025.

    Modalities
    text, image, video → text
    Context
    125K context
    Parameters
    MoE: 16B total, 3B active
    License
    MIT
    At launch
    Open weights
    • Released with a tech blog and cookbook on Hugging Face
    • Same 16B-total, 3B-active mixture-of-experts design as Kimi-VL
    • Weights under the MIT License
    Source: Kimi-VL (GitHub)
  2. CodingOpen weights

    Kimi-Dev-72B

    Kimi-Dev-72B is Moonshot AI's open-weights coding model for software engineering issue resolution. It was optimized with large-scale reinforcement learning in which it patches real repositories in Docker and is rewarded only when the full test suite passes.

    Modalities
    text → text
    Parameters
    72B
    License
    MIT
    At launch
    Open weights
    • Targets SWE-bench-style repository issue resolution
    • Available on Hugging Face and GitHub
    • Weights under the MIT License
    Source: moonshotai/Kimi-Dev-72B (Hugging Face)

April 2025

2 releases
  1. SpeechOpen weights

    Kimi-Audio

    Kimi-Audio is an open-source audio foundation model for audio understanding, generation, and conversation in one framework. It handles tasks such as speech recognition, audio question answering, captioning, emotion recognition, and end-to-end speech conversation.

    Modalities
    text, audio → text, audio
    Parameters
    7B
    License
    MIT
    At launch
    Open weights
    • Pre-trained on more than 13 million hours of audio and text data
    • Hybrid input of continuous acoustic and discrete semantic tokens
    • Kimi-Audio-7B-Instruct released April 25, 2025; base Kimi-Audio-7B weights followed April 27
    Source: Kimi-Audio (GitHub)
  2. MultimodalOpen weights

    Kimi-VL

    Kimi-VL is an efficient open-weights mixture-of-experts vision-language model for multimodal reasoning, long-context understanding, and agent tasks. It was released as Kimi-VL-A3B-Instruct and Kimi-VL-A3B-Thinking.

    Modalities
    text, image, video → text
    Context
    125K context
    Parameters
    MoE: 16B total, 3B active
    License
    MIT
    At launch
    Open weights
    • Activates about 2.8B parameters in its language decoder
    • 128K extended context window
    • Native-resolution MoonViT vision encoder for high-resolution inputs
    Source: moonshotai/Kimi-VL-A3B-Instruct (Hugging Face)

February 2025

1 release
  1. Small modelOpen weights

    Moonlight

    Moonlight is a 3B-active, 16B-total mixture-of-experts model trained on 5.7T tokens with the Muon optimizer. Moonshot released it to show that Muon scales to large-model training without hyperparameter tuning.

    Modalities
    text → text
    Parameters
    MoE: 16B total, 3B active
    License
    MIT
    At launch
    Open weights
    • Moonshot reports Muon is about 2 times more sample efficient than Adam
    • Released pretrained, instruction-tuned, and intermediate checkpoints
    • Muon implementation open-sourced alongside the model
    Source: Moonlight: Muon is Scalable for LLM Training (GitHub)

January 2025

1 release
  1. ReasoningProprietary

    Kimi k1.5

    Kimi k1.5 is Moonshot AI's multimodal reasoning model trained with reinforcement learning, which Moonshot describes as o1-level. Its training scaled the reinforcement learning context window to 128K and used partial rollouts for efficiency.

    Modalities
    text, image → text
    At launch
    Consumer app
    • Jointly trained on text and vision data
    • Reports long chain-of-thought and short chain-of-thought results
    • Moonshot said the service would come to kimi.ai soon after the January 20, 2025 report
    Source: Kimi k1.5: Scaling Reinforcement Learning with LLMs (GitHub)

Cite and reuse this data

These release records are free to reuse under CC BY 4.0, including commercially. Credit The AI Engineering and link to this page's URL.

Dates are the labs' official announcement dates, and every record links to its source. List fields in the CSV are separated by " | ".

Citation

The AI Engineering Editors. "Moonshot AI models: release history and timeline." The AI Engineering, updated Oct 24, 2026. https://theaiengineering.com/releases/moonshot-ai

HTML link

<a href="https://theaiengineering.com/releases/moonshot-ai">Moonshot AI models: release history and timeline</a> (The AI Engineering)
Full dataset
1,167 releases from 55 labs
Coverage
to
Updated
License
CC BY 4.0
Publisher
The AI Engineering

Other labs