Kimi K3
kimi-k3Kimi K3 is a 2.8T parameter 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... Built by MoonshotAI.
Prices updated
Input price
$3.00
per 1M tokens · Standard
Output price
$12
per 1M tokens · Standard
Input limit
1M
tokens
Output limit
944K
tokens
Input formats
Output formats
Kimi K3 price history
6 price records since 9/25/2026 · 4 changes
Kimi K3 cost calculator
$122/month
Standard pricing
Overview
What is Kimi K3?
Kimi K3 is a text & reasoning and vision model from MoonshotAI. Kimi K3 is a 2.8T parameter 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... Its 1M context window and $3.00 input price make it a candidate for quality-focused production applications.
Benchmarks
Kimi K3 benchmarks & speed
How Kimi K3 scores on standardized evaluations, and where it lands among every model we track.
43.6
Intelligence Index
45tok/s
Output speed
1368
DesignArena Elo
Reasoning
GPQA Diamond
93.5%
Graduate-level scientific reasoning · top 7%
HLE
46.9%
Humanity's Last Exam · top 9%
Coding
SciCode
59.5%
Python for scientific computing · top 6%
Latency & design
- Time to first token
- 3.83s
- DesignArena win rate
- 63.1%
- Design battles judged
- 7,980
Independent scores from Artificial Analysis and DesignArena · updated 10/8/2026. Higher is better; ranks compare against every model we track with that score.
Rates
Kimi K3 pricing
Pricing is based on token usage. Provider-specific caching, batch, regional, and tool charges may affect the final cost.
Input · Standard
$3.00
Per 1M tokens
Cached input · Standard
$0.43
Per 1M tokens
Capabilities
Kimi K3 Tools
Tools available when using Kimi K3 through supported provider APIs.
Function calling
Supported
Structured outputs
Supported
JSON mode
Supported
Reasoning
Supported
Built-in web search
Not supported
Log probabilities
Supported
Deterministic seed
Supported
Parallel tool calls
Not supported
Prompt caching
Supported
Strengths and limitations
Strengths
- Measured results (AA Intelligence 43.6 · DesignArena Elo 1368) position it for demanding production work.
- 1M context supports large documents, repositories and extended conversations.
- Supports text & reasoning and vision workloads in one model.
Limitations
- Generated tokens cost 4× more than input tokens, which matters for verbose responses.
- Large context capacity does not guarantee consistent retrieval across the entire prompt.
- Arena Elo and MMLU-Pro are directional; test accuracy, latency and reliability on your own workload before committing.
Arena Elo, MMLU-Pro and pricing figures are illustrative; validate current vendor terms before purchase.
Same provider