Moonshot AI (Kimi)

Kimi K3

Type
language model
Context
1,049K tokens
Max output
131K
Released
27 July 2026
API string
accounts/fireworks/models/kimi-k3
Weights
open

Prices by provider

Provider Input, $ per 1M Output, $ per 1M In our data since
crof $2 $8 31 Jul 2026
DeepInfra $2.7 $13.5 31 Jul 2026
opencode-go $3 $15 31 Jul 2026
cline-pass $3 $15 31 Jul 2026
aiand $3 $12.5 31 Jul 2026
moonshotai-cn $3 $15 31 Jul 2026
berget $3 $15 31 Jul 2026
synthetic $3 $15 31 Jul 2026
novita-ai $3 $15 31 Jul 2026
perplexity-agent $3 $15 31 Jul 2026
AIHubMix $3 $15 31 Jul 2026
Baseten $3 $15 31 Jul 2026
Cloudflare AI Gateway $3 $15 31 Jul 2026
Cortecs $3 $15 31 Jul 2026
CrossModel $3 $15 31 Jul 2026
EmpirioLabs AI $3 $15 31 Jul 2026
Fireworks AI $3 $15 31 Jul 2026
Hugging Face $3 $15 31 Jul 2026
LLM Gateway $3 $15 31 Jul 2026
Modal $3 $15 2 Aug 2026
Moonshot AI $3 $15 31 Jul 2026
Nebius Token Factory $3 $15 31 Jul 2026
OpenCode Zen $3 $15 31 Jul 2026
OpenRouter $3 $15 31 Jul 2026
Together AI $3 $15 31 Jul 2026
Vercel $3 $15 31 Jul 2026
Weights & Biases $3 $15 31 Jul 2026
ZenMux $3 $15 31 Jul 2026
hyper $3.266 $16.332 31 Jul 2026
Venice AI $3.75 $18.75 31 Jul 2026

Only the base tier and only per-token prices are shown. Batch, discounted and cached rates, as well as prices per image or per second of video, do not go into this table: they cannot stand in the same column as a price per million tokens.

The date in the last column is the day this price first entered our collection. The price may well be older: before that day we simply were not recording it. It has not changed since — otherwise a new row with a new date would stand in its place.

Measurement results

Task set Result Run conditions Measured by
Arena Score, web development 1,675.56 1,664–1,687 effort: max
Arena Score, programming 1,499.63 1,480–1,519 effort: max
Arena Score, expert questions 1,495.53 1,462–1,529 effort: max
Arena Score, long queries 1,492.83 1,477–1,509 effort: max
Arena Score, multi-turn dialogue 1,485.52 1,460–1,511 effort: max
Arena Score, hard prompts 1,484.46 1,472–1,497 effort: max
Arena Score in Russian 1,477.9 1,448–1,508 effort: max
Arena Score in English 1,477.38 1,461–1,493 effort: max
Arena Score, instruction following 1,473.93 1,456–1,492 effort: max
Arena Score, overall 1,473.33 1,463–1,483 effort: max
Arena Score, creative writing 1,464.37 1,440–1,489 effort: max
Mock AIME 2024–2025 — olympiad problems 97.22 % effort: max · with a tuned harness Epoch evaluations
GPQA Diamond — graduate-level questions 90.83 % effort: max · with a tuned harness Epoch evaluations
WeirdML — unusual machine learning tasks 82.57 % effort: max https://htihle.github.io/weirdml.html
FrontierMath, levels 1–3 72.18 % effort: max Epoch evaluations
FrontierCode — patches fit to be merged into a project 44.2 % https://cognition.com/frontiercode
SimpleQA Verified — factual accuracy 42.7 % effort: max Epoch evaluations
APEX-Agents 39.3 %
FrontierMath, level 4 — research-grade problems 39.02 % effort: max Epoch evaluations
Chess puzzles 35.82 % effort: max Epoch evaluations
CritPt — physics problems 23.4 % effort: max
Arena Score, agent tasks 0.1 0–0 effort: max