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 4 CarbonGemini 4 ArgonGemini 3.1 Pro PreviewChat with these 3
AttributeGemini 4 Carbongemini/gemini-4-carbon50% offGemini 4 Argongemini/gemini-4-argon50% offGemini 3.1 Pro Previewgemini/gemini-3.1-pro-preview
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$ 4.00$ 2.00$ 4.00$ 2.00$ 2.00>= 200K$ 4.00
Output$ 20.00$ 10.00$ 20.00$ 10.00$ 12.00>= 200K$ 18.00
Cache read$ 0.20$ 0.10$ 0.20$ 0.10$ 0.20>= 200K$ 0.40
Cache write$ 4.00$ 2.00$ 4.00$ 2.00$ 2.00>= 200K$ 4.00
What it is good at
Best for
  • Deep debugging and agentic coding
  • High-precision knowledge work
  • Mathematical and scientific reasoning
  • Open-ended research
  • Agentic software engineering
  • Mathematical and scientific reasoning
  • Multimodal document analysis
  • Specialist knowledge work
  • Advanced coding agents
  • Large multimodal corpora
Published benchmarks
Terminal-Bench 2.0pass_rate——
  • 68.5%effort high · Terminus 2(setup not fully disclosed)
Terminal-Bench 4.0pass_rate—
  • 57.4%effort high(setup not fully disclosed)
—
BrowseComp (with tools)accuracy——
  • 85.9%effort high · tools: browse, python, search(setup not fully disclosed)
Agents' Last Exam v1.1score
  • 77.4%effort high · ALE-Claw(setup not fully disclosed)
  • 69%effort high · ALE-Claw(setup not fully disclosed)
—
DeepSWE 1.1resolved_rate—
  • 77.9%effort high · mini-swe(setup not fully disclosed)
—
SWE-Bench Verifiedresolved_rate——
  • 80.6%effort high(setup not fully disclosed)
SWE-sweepscore
  • 9.6%effort high · mini-swe-agent(setup not fully disclosed)
  • 4.3%effort high · mini-swe-agent(setup not fully disclosed)
—
Agents' Last Exam (ALE-Claw, binary pass)pass_rate—
  • 39.5%effort high · ALE-Claw(setup not fully disclosed)
—
OSWorld 2.0 (v2026.08.08 offline, partial score)score—
  • 69.2%effort high · Gemini CUA · 3 runs · Compaction following Claude Opus 5.5 system card(setup not fully disclosed)
—
OSWorld 2.1 (offline subset, partial score)score
  • 81.5%effort high · Gemini CUA · 3 runs · Compaction following Claude Opus 5.5 system card(setup not fully disclosed)
  • 74.7%effort high · Gemini CUA · 3 runs · Compaction following Claude Opus 5.5 system card(setup not fully disclosed)
—
GraphWalks BFS 256K to 1Mf1
  • 96.2%effort high(setup not fully disclosed)
  • 84.2%effort high(setup not fully disclosed)
—
GraphWalks BFS up to 128Kf1—
  • 99.7%effort high(setup not fully disclosed)
—
PostTrainBench 1.1score—
  • 45.3%effort high · OpenCode(setup not fully disclosed)
—
Humanity's Last Examaccuracy——
  • 44.4%effort high(setup not fully disclosed)
GPQA Diamondaccuracy——
  • 94.3%effort high(setup not fully disclosed)
LABBench 2score—
  • 88.8%effort high(setup not fully disclosed)
—
Terminal-Bench-Science 0.1accuracy—
  • 57.6%effort high(setup not fully disclosed)
—
LVBench (static)accuracy—
  • 91.7%effort high(setup not fully disclosed)
—