CrossModel
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.

Gemini 3.7 FlashGemini 3.6 FlashGemini 3.5 Flash LiteChat with these 3
AttributeGemini 3.7 Flashgemini/gemini-3.7-flashGemini 3.6 Flashgemini/gemini-3.6-flashGemini 3.5 Flash Litegemini/gemini-3.5-flash-lite
Specs
Context window1M1M1M
Max output66K66K66K
Input modalitiesTextImageAudioVideoTextImageAudioVideoTextImageAudioVideo
Output modalitiesTextTextText
Capabilities
StreamingSupportedSupportedSupported
Tool callingSupportedSupportedSupported
Structured output (JSON)SupportedSupportedSupported
Image inputSupportedSupportedSupported
Thinking modeSupportedSupportedSupported
Price / 1M tokens
Input$ 0.75$ 1.50$ 0.30
Output$ 3.75$ 7.50$ 2.50
Cache read$ 0.075$ 0.15$ 0.030
Cache write$ 0.75$ 1.50$ 0.30
What it is good at
Best for
  • Complex coding agents
  • Design-to-code fidelity
  • Long multimodal workflows
  • Rapid agentic coding loops
  • Efficient knowledge work
  • Spatial and multimodal reasoning
  • Subagents and automation
  • Document parsing
Published benchmarks
FrontierCode v1.1 (Main)score
  • 43.6%(setup not fully disclosed)
——
Terminal-Bench 2.1pass_ratesetups differ by: harness
  • 85.8%(setup not fully disclosed)
  • 78%Terminus 2(setup not fully disclosed)
  • 54%Terminus 2(setup not fully disclosed)
SWE-Bench Pro (Public)pass_rate—
  • 58.7%(setup not fully disclosed)
  • 54.2%(setup not fully disclosed)
OSWorld 2.0success_rate
  • 47.9%(setup not fully disclosed)
——
OSWorld-Verifiedsuccess_rate—
  • 83%(setup not fully disclosed)
  • 74%(setup not fully disclosed)
GDPval-AA v2elo
  • 1525(setup not fully disclosed)
  • 1421(setup not fully disclosed)
  • 1140(setup not fully disclosed)
GDM-MRCR v2 (128K, 8-needle)accuracy
  • 97%(setup not fully disclosed)
——
CharXiv Reasoning (no tools)accuracy
  • 84.5%(setup not fully disclosed)
  • 85.2%(setup not fully disclosed)
  • 74.5%(setup not fully disclosed)
HLE-Verifiedaccuracy
  • 53.6%(setup not fully disclosed)
——