Kimi K3 Benchmarks, Pricing & Rankings
Moonshot AI · 2026-07-30 · Open · Current
Snapshot 2026-10-14 · updated weekly
Kimi K3 ranks #17 of 42 on the AMR leaderboard.
Kimi K3 benchmark scores
| GPQA Diamond | 93.5% | |
| MMLU-Pro | 84.6% | |
| SWE-Bench Verified | 66.8% | |
| AIME 2025 | 88.2% | |
| LiveCodeBench v6 | 74.9% | |
| Arena Elo | 1408 |
Kimi K3 price, speed & context
Input price
$0.50
per 1M tokens
Output price
$2
per 1M tokens
Speed
98
tok/s
Context
512K
tokens
About Kimi K3
Kimi K3 is the best open-weights model by GPQA (93.5%) and the strongest open reasoning model you can self-host today, from Moonshot AI.
Models around Kimi K3 in the rankings
More Moonshot AI models
Kimi K3 FAQ
Where does Kimi K3 rank?
Kimi K3 ranks #17 of 42 tracked AI models on the AMR aggregate as of October 2026, with an aggregate score of 87.5. Its strongest dimension is GPQA Diamond at 93.5%.
How much does Kimi K3 cost?
Kimi K3 costs $0.50 per million input tokens and $2 per million output tokens on the standard API. At a typical 1:3 input/output agent workload, that blends to roughly $1.63 per million tokens.
What is the Kimi K3 context window?
Kimi K3 supports a 512K token context window and generates at a median 98 output tokens per second. That context fits roughly 341 standard documents in a single prompt.
Is Kimi K3 open source?
Yes — Kimi K3 ships open weights that you can download and self-host or serve through any compatible inference provider.
What is better than Kimi K3?
Claude Haiku 5.5 ranks one place higher at #16 with an AMR score of 87.9. Kimi K3 still wins 6 of the tracked dimensions head-to-head — check the comparison table to see exactly which.