Kimi K3
by Moonshotai
Kimi K3 is an ultra-large-scale, open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and i
Models
Kimi K3
Added 16 July 2026
Overview
Kimi K3 is an ultra-large-scale, open-weight multimodal reasoning model from Moonshot AI. It is suited for complex coding, knowledge work, and long-horizon agentic workflows, and is particularly strong at navigating...
How to use / API access
Call via OpenRouter with model id `moonshotai/kimi-k3`.
- openrouter ·
moonshotai/kimi-k3Docs
OpenRouter id: moonshotai/kimi-k3
Benchmarks
- Intelligence Index (artificial-analysis) : 43.6 index
- Coding Index (artificial-analysis) : 76.2 index
- Agentic Index (artificial-analysis) : 50 index
- Design Arena agenticgamedev (other) : 1252 elo · #3
- Design Arena androidnative (other) : 1258 elo · #5
- Design Arena fullstack (other) : 1331 elo · #2
- Design Arena godotgamedev (other) : 1199 elo · #10
Notes
Kimi K3 is listed in the EDNA Models directory from the OpenRouter catalogue.
In the AI Pulse
Recent daily briefs from the Enterprise DNA AI Pulse that mention this. The short daily read on what is actually happening in AI.
- Under the Radar2 Sept 2026
Harness choice swings agent cost 17.5x on the same task
- Frontier Labs Watch10 Aug 2026
Meta ships a 30B open-weight agentic model built to run on a single consumer GPU, and Zuckerberg publishes an essay attacking closed AI labs
- AI Trends Pulse9 Aug 2026
DeepSeek-V4-Flash and MiniMax-H3 are dominating Hugging Face's trending list this week
- AI Trends Pulse7 Aug 2026
Qwen3.8 Max took #1 on Artificial Analysis' agentic capability index (526 points on
- Under the Radar1 Aug 2026
Free 2.8T-param Kimi K3 access via TokenRouter is spreading through an X swarm, not one viral post.
- AI Trends Pulse1 Aug 2026
Kimi K3 (Moonshot, 2.8T params) is the top-trending model on Hugging Face this week
- Under the Radar31 July 2026
Extreme local inference of frontier-scale open models keeps escalating.
- Under the Radar30 July 2026
Harness choice, not just model choice, swings agentic task cost up to 6x on identical work.
Get the free Developer’s Field Guide
A 27-page field guide to the AI coding workflow with Claude. Claude Code, MCP servers, the prompt patterns that work, and what to delegate. Free.
Enter your work email. We send it straight over, plus a few short notes worth knowing. Unsubscribe any time.