Model details
Kimi Latest
The model overview is temporarily unavailable.
Quick Info
Powered by- Provider
- Fireworks AI
- Model key
- accounts/fireworks/routers/kimi-latest
- Release date
- Jul 27, 2026
- Last updated
- Sep 15, 2026
- Input modalities
- Output modalities
- Capabilities
Cost
- Input token cost
- $3.00
- Output token cost
- $15.00
Limits
- Output tokens
- 131,072 tokens
- Context window
- 1,048,576 tokens
OpenCode
Model variants
Latest news about Kimi Latest
Fireworks AI
Kimi K3 went live on July 16, 2026 through Moonshot's API at kimi.com, the Kimi mobile apps, Kimi Work, and Kimi Code, with OpenRouter providing immediate access for users without a Moonshot account. The model is a 2.8-trillion-parameter multimodal system with native image and video input, a 1,048,576-token context win Independent benchmark results placed Kimi K3 fourth on the Artificial Analysis Intelligence Index with a score of 57 out of 189 models, behind Claude Fable 5 and two GPT-5.6 Sol reasoning settings but ahead of Claude Opus 4.8, GPT-5.5 at xhigh, Claude Sonnet 5, and GLM-5.2; the model also topped the Arena.ai Code WebDe
Fireworks AI
Moonshot AI released Kimi K3 on July 16, 2026, a 2.8-trillion-parameter open-weight model that the article describes as the largest open-weight AI system ever announced. The piece emphasizes that Kimi K3 launched at flagship pricing—$3 per million input tokens and $15 per million output tokens—a deliberate break from t The deep dive benchmarks Kimi K3 against Claude Fable 5, GPT-5.6, and other frontier models, and it examines the economics of what happens when frontier-class capability ships with downloadable weights and a permissive license. An August 5 update corrects earlier claims that weights would ship under a Modified MIT lice
Fireworks AI
Moonshot AI's Kimi K3 debuted at the top of Arena's Frontend Code leaderboard, a result the MarketWatch coverage framed as evidence that the capability gap between U.S. and Chinese AI labs could be closing faster than expected. The 2.8-trillion-parameter model was announced as the world's largest open model, with weigh The launch contributed to a 1.4% drop in the Nasdaq Composite on Friday as markets weighed the implications of a Chinese model achieving frontier-class results with less compute than U.S. labs. Analysts drew parallels to the DeepSeek moment, noting that open models catching up in capability would put pricing pressure o