Back to Models
moonshotai logo
moonshotai/kimi-k2-0905
Not Available

Kimi K2 0905

Kimi K2 0905 is the September update of [Kimi K2 0711](moonshotai/kimi-k2). It is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32 billion active per forward pass. It supports long-context inference up to 256k tokens, extended from the previous 128k. This update improves agentic coding with higher accuracy and better generalization across scaffolds, and enhances frontend coding with more aesthetic and functional outputs for web, 3D, and related tasks. Kimi K2 is optimized for agentic capabilities, including advanced tool use, reasoning, and code synthesis. It excels across coding (LiveCodeBench, SWE-bench), reasoning (ZebraLogic, GPQA), and tool-use (Tau2, AceBench) benchmarks. The model is trained with a novel stack incorporating the MuonClip optimizer for stable large-scale MoE training.

9/4/2025
131,072 tokens
#92 Text (Japanese)

Specifications

Modalities

Input
text
Output
text

Supported Parameters

frequency_penalty
logit_bias
logprobs
max_tokens
min_p
presence_penalty
repetition_penalty
response_format
seed
stop
structured_outputs
temperature
tool_choice
tools
top_k
top_logprobs
top_p

Frequently asked questions

Is Kimi K2 0905 available on FastMetal?
Not at the moment. Kimi K3, from the same lab, is available on the FastMetal API today.
What is the context window of Kimi K2 0905?
131,072 tokens, shared between the prompt and the response.
How does Kimi K2 0905 rank?
#92 on the public arena's Japanese board (ELO 1,378). Ranks move as the leaderboard is updated.

Leaderboard

Text
OverallELO: 1,418
#118
JapaneseELO: 1,378
#92
ChineseELO: 1,457
#112
KoreanELO: 1,365
#103
EnglishELO: 1,420
#137
frenchELO: 1,421
#129
germanELO: 1,422
#92
spanishELO: 1,421
#95
russianELO: 1,423
#100
CodingELO: 1,467
#114
MathELO: 1,415
#120
Creative WritingELO: 1,380
#129
Instruction FollowingELO: 1,388
#146
Hard PromptsELO: 1,436
#121
Multi-TurnELO: 1,404
#142