Claude Fable 5$22.000/MClaude Opus 5$11.000/MClaude Opus 4.8$11.000/MClaude Opus 4.7$11.000/MClaude Opus 4.6$11.000/MClaude Opus 4.5$33.000/MClaude Sonnet 3.7$6.600/MClaude Opus 3$33.000/MClaude 2.1$12.800/MClaude 2$12.800/MGPT-5.6 Sol$12.500/MGPT-5.6 Terra$5.000/MGPT-5.5$12.500/MGPT-5.2$5.425/MGPT-5.2-Codex$5.425/MGPT-5$3.875/MGPT-4.5$97.500/MGPT-4 Turbo Preview$16.000/MGPT-4$39.000/MGPT-4-32k$78.000/Mo3$19.000/Mo3-mini$2.090/Mo4-mini$2.090/Mo1$28.500/Mo1-mini$5.700/Mo1-preview$28.500/MGemini 3.5 Pro$5.000/MGemini 3.1 Pro$5.000/MGemini 3 Pro$5.000/MGemini 2.5 Pro$3.875/MClaude Fable 5$22.000/MClaude Opus 5$11.000/MClaude Opus 4.8$11.000/MClaude Opus 4.7$11.000/MClaude Opus 4.6$11.000/MClaude Opus 4.5$33.000/MClaude Sonnet 3.7$6.600/MClaude Opus 3$33.000/MClaude 2.1$12.800/MClaude 2$12.800/MGPT-5.6 Sol$12.500/MGPT-5.6 Terra$5.000/MGPT-5.5$12.500/MGPT-5.2$5.425/MGPT-5.2-Codex$5.425/MGPT-5$3.875/MGPT-4.5$97.500/MGPT-4 Turbo Preview$16.000/MGPT-4$39.000/MGPT-4-32k$78.000/Mo3$19.000/Mo3-mini$2.090/Mo4-mini$2.090/Mo1$28.500/Mo1-mini$5.700/Mo1-preview$28.500/MGemini 3.5 Pro$5.000/MGemini 3.1 Pro$5.000/MGemini 3 Pro$5.000/MGemini 2.5 Pro$3.875/M
Moonshot AI
Moonshot AI
Frontier

Kimi K3

Frontier

Moonshot AI's 2.7-trillion-parameter flagship, released in July 2026 as the largest open-weight model published to date and the first Chinese model widely treated as competitive with the top US frontier systems. It runs an always-on reasoning mode Moonshot calls "thinking mode", handles a 1M-token context window, and understands images natively. Two variants shipped: K3 Max for chat and agent work, K3 Swarm Max for large-scale parallel processing.

Kimi K3 is a frontier AI model from Moonshot AI. It costs $3.000 per million input tokens and $15.000 per million output tokens (blended $6.600/M), with a 1,049,000-token context window.

INPUT
$3.000/M
per million input tokens
OUTPUT
$15.000/M
per million output tokens
CONTEXT
1,049,000
tokens
What it is good at
  • Largest open-weight model released to date (2.7T parameters)
  • Always-on "thinking mode" reasoning
  • 1M-token context
  • Native visual understanding
  • Strong competitive coding performance
Typical use cases
  • Frontier-class reasoning without US-vendor lock-in
  • Self-hosted deployments where weights matter
  • Agentic coding
  • Long-context analysis

Benchmarks

No published benchmark scores tracked for this model yet. Frontier and reasoning models from major providers have scores; smaller models and inference-host variants typically inherit the underlying open-weights score.

More from Moonshot AI

See all 5

Frequently asked questions

How much does Kimi K3 cost?

Kimi K3 costs $3.000 per million input tokens and $15.000 per million output tokens, for a blended reference rate of $6.600 per million tokens.

What is Kimi K3's context window?

Kimi K3 supports up to 1,049,000 tokens of context in a single request.

What is Kimi K3 best for?

Kimi K3 is well suited to Largest open-weight model released to date (2.7T parameters), Always-on "thinking mode" reasoning and 1M-token context.

Who makes Kimi K3?

Kimi K3 is developed and served by Moonshot AI. It was released in Jul 2026.

Compared with

Terms used on this page