Kimi K2 Instruct

Overview

Kimi K2 is a state-of-the-art mixture-of-experts (MoE) language model with 32 billion activated parameters and 1 trillion total parameters. Trained with the MuonClip optimizer, it achieves exceptional performance across frontier knowledge, reasoning, and coding tasks while being meticulously optimized for agentic capabilities. The instruct variant is post-trained for drop-in, general-purpose chat and agentic experiences without long thinking.

Kimi K2 Instruct was released on July 11, 2025. API access is available through Fireworks, Novita.

Benchmarks

Kimi K2 Instruct Performance Across Datasets

Scores sourced from the model's scorecard, paper, or official blog posts

llm-stats.com - Sat Dec 06 2025

Pricing

Pricing, performance, and capabilities for Kimi K2 Instruct across different providers:

Provider	Input ($/M)	Output ($/M)	Max Input	Max Output	Latency (s)	Throughput	Quantization	Input	Output
Fireworks	$0.50	$0.50	200.0K	200.0K	—	—	—	Text Image Audio Video	Text Image Audio Video
Novitafp8	$0.57	$2.30	131.1K	131.1K	0.95	45.0 tok/s	fp8	Text Image Audio Video	Text Image Audio Video

Price Comparison for Kimi K2 Instruct

Price per 1M input tokens (USD), lower is better

llm-stats.com - Sat Dec 06 2025

Throughput Comparison for Kimi K2 Instruct

Tokens per second, higher is better

llm-stats.com - Sat Dec 06 2025

Latency Comparison for Kimi K2 Instruct

Time to first token (s), lower is better

llm-stats.com - Sat Dec 06 2025

Example Outputs

API Access