# Kimi: what I've logged

> Kimi comes up in 8 posts of my AI log, from September 2025 to July 2026, and each item below links to the primary source. The sources come from arXiv and Moonshot AI, filed under robotics, open weights and image, video, world.

## Items

1. **Kimi K3: Open Frontier Intelligence** (arXiv, 27 Jul 2026): The technical report for Kimi K3, covering the 2.8T mixture-of-experts design, Kimi Delta Attention and the 1M-token context. [Primary source](https://arxiv.org/abs/2607.24653) · [Post](https://heyaliasif.com/ai/2026-07-27-week-in-ai#item-6)
2. **Kimi K3** (Moonshot AI, 13 Jul 2026): A 2.8-trillion-parameter mixture-of-experts model with a 1M-token context, native vision and open weights under a conditional-commercial license. [Primary source](https://huggingface.co/moonshotai/Kimi-K3) · [Post](https://heyaliasif.com/ai/2026-07-13-week-in-ai#item-2)
3. **Kimi K2.7 Code** (Moonshot AI, 8 Jun 2026): An open coding-focused follow-up to K2.6 under a modified MIT license, released June 12. [Primary source](https://huggingface.co/moonshotai/Kimi-K2.7-Code) · [Post](https://heyaliasif.com/ai/2026-06-08-week-in-ai#item-2)
4. **Kimi K2.6** (Moonshot AI, 20 Apr 2026): A 1T-parameter mixture-of-experts model with 32B active, built for long coding runs and agent swarms of up to 300 parallel sub-agents. [Primary source](https://www.kimi.com/blog/kimi-k2-6) · [Post](https://heyaliasif.com/ai/2026-04-20-week-in-ai#item-3)
5. **Kimi K2.5** (Moonshot AI, 26 Jan 2026): A natively multimodal open model with an agent-swarm mode, from the lab behind Kimi K2. [Primary source](https://www.kimi.com/blog/kimi-k2-5) · [Post](https://heyaliasif.com/ai/2026-01-26-week-in-ai#item-1)
6. **Kimi K2 Thinking** (Moonshot AI, 3 Nov 2025): An open reasoning model that Moonshot says can chain 200 to 300 tool calls in a row without drifting. [Primary source](https://huggingface.co/moonshotai/Kimi-K2-Thinking) · [Post](https://heyaliasif.com/ai/2025-11-03-week-in-ai#item-1)
7. **Kimi Linear 48B-A3B** (Moonshot AI, 27 Oct 2025): A hybrid linear-attention design that cuts the memory cache sharply and speeds up decoding on long inputs. [Primary source](https://huggingface.co/moonshotai/Kimi-Linear-48B-A3B-Instruct) · [Post](https://heyaliasif.com/ai/2025-10-27-week-in-ai#item-3)
8. **Kimi K2 Instruct 0905** (Moonshot AI, 1 Sept 2025): Context goes from 128K to 256K tokens and the agentic coding behavior is tuned, still as open weights. [Primary source](https://huggingface.co/moonshotai/Kimi-K2-Instruct-0905) · [Post](https://heyaliasif.com/ai/2025-09-01-week-in-ai#item-1)

## Posts

- 2026-07-27 [Gemini Robotics 2, plus Korean labs open 700B-class models](https://heyaliasif.com/ai/2026-07-27-week-in-ai)
- 2026-07-13 [Grok 4.5 lands, Kimi K3 and Inkling open up huge models](https://heyaliasif.com/ai/2026-07-13-week-in-ai)
- 2026-06-08 [Claude Fable 5 and Mythos 5 arrive, plus Kimi K2.7 Code](https://heyaliasif.com/ai/2026-06-08-week-in-ai)
- 2026-04-20 [GPT-5.5, DeepSeek V4 preview and Kimi K2.6 in one week](https://heyaliasif.com/ai/2026-04-20-week-in-ai)
- 2026-01-26 [Kimi K2.5, DeepMind's Project Genie, and the OpenClaw frenzy](https://heyaliasif.com/ai/2026-01-26-week-in-ai)
- 2025-11-03 [Kimi K2 Thinking, Google's space-compute idea, and Nested Learning](https://heyaliasif.com/ai/2025-11-03-week-in-ai)
- 2025-10-27 [MiniMax-M2 for agents, Cursor 2.0, and a robot you can preorder](https://heyaliasif.com/ai/2025-10-27-week-in-ai)
- 2025-09-01 [Kimi K2 stretches to 256K, Switzerland opens Apertus](https://heyaliasif.com/ai/2025-09-01-week-in-ai)

Canonical: https://heyaliasif.com/ai/topics/kimi