Back to model catalog

Compare models

Context, limits, capabilities, price and use cases side by side — up to 4 models at a time. The URL holds your selection, so you can share it.

Kimi K3.1 ProKimi K3Chat with these 2
AttributeKimi K3.1 Promoonshot/kimi-k3.1-proKimi K3moonshot/kimi-k3
Specs
Context window1M1M
Max output1M1M
Input modalitiesTextImageVideoTextImageVideo
Output modalitiesTextText
Capabilities
StreamingSupportedSupported
Tool callingSupportedSupported
Structured output (JSON)SupportedSupported
Image inputSupportedSupported
Thinking modeSupportedSupported
Price / 1M tokens
Input$ 2.00$ 3.00
Output$ 10.00$ 15.00
Cache read$ 0.20$ 0.30
Cache write$ 2.00$ 3.00
What it is good at
Best for
  • Coding agents
  • 3D modeling and CAD
  • Financial research
  • Long-horizon engineering
  • Deep research
  • Computer use
Published benchmarks
Terminal-Bench 2.1pass_rate—
  • 88.3%(setup not fully disclosed)
Terminal-Bench 4.0pass_rate
  • 65.7%effort max · Claude Code(setup not fully disclosed)
—
BrowseComp (with tools)accuracysetups differ by: context_management, reasoning, temperature, top_p
  • 92.1%effort max · compaction at 200K tokens(setup not fully disclosed)
  • 91.2%(setup not fully disclosed)
DeepSearchQAf1—
  • 95%(setup not fully disclosed)
DeepSWEresolved_rate—
  • 67.5%(setup not fully disclosed)
DeepSWE 1.1resolved_rate
  • 75.5%effort max · Claude Code(setup not fully disclosed)
—
ProgramBenchscore—
  • 77.8%(setup not fully disclosed)
OSWorld 2.1 (partial credit)score
  • 79.6%effort max · Kimi Agent(setup not fully disclosed)
—
OSWorld Verifiedsuccess_rate—
  • 84.8%(setup not fully disclosed)
OfficeQA Proscore
  • 67.4%effort max · Claude Code(setup not fully disclosed)
—
MMMU Pro (no tools)accuracy
  • 84.1%effort max · 4 runs(setup not fully disclosed)
—
Humanity's Last Exam Full (with tools)accuracysetups differ by: temperature, top_p
  • 70%effort max(setup not fully disclosed)
  • 56%effort max(setup not fully disclosed)
GPQA Diamondaccuracysetups differ by: temperature, top_p
  • 94.2%effort max(setup not fully disclosed)
  • 93.5%effort max(setup not fully disclosed)
MCPMark Verifiedscore—
  • 94.5%(setup not fully disclosed)