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.

GPT-5.6 SolGPT-5.6 TerraGPT-5.5GPT-6 AstraChat with these 4
AttributeGPT-5.6 Solopenai/gpt-5.6-solPromo priceGPT-5.6 Terraopenai/gpt-5.6-terraGPT-5.5openai/gpt-5.5GPT-6 Astraopenai/gpt-6-astra
Specs
Context window1M1M1M1M
Max output128K128K128K128K
Input modalitiesTextImageTextImageTextImageTextImage
Output modalitiesTextTextTextText
Capabilities
StreamingSupportedSupportedSupportedSupported
Tool callingSupportedSupportedSupportedSupported
Structured output (JSON)SupportedSupportedSupportedSupported
Image inputSupportedSupportedSupportedSupported
Thinking modeSupportedSupportedSupportedSupported
Price / 1M tokens
Input$ 5.00$ 4.00>= 272K$ 10.00$ 8.00$ 2.00>= 272K$ 4.00$ 5.00>= 272K$ 10.00$ 10.00>= 272K$ 20.00
Output$ 30.00$ 20.00>= 272K$ 45.00$ 30.00$ 12.00>= 272K$ 18.00$ 30.00>= 272K$ 45.00$ 50.00>= 272K$ 75.00
Cache read$ 0.50$ 0.40>= 272K$ 1.00$ 0.80$ 0.20>= 272K$ 0.40$ 0.50>= 272K$ 1.00$ 1.00>= 272K$ 2.00
Cache write$ 6.25$ 5.00>= 272K$ 12.50$ 10.00$ 2.50>= 272K$ 5.00$ 5.00>= 272K$ 10.00$ 12.50>= 272K$ 25.00
What it is good at
Best for
  • Frontier agentic coding
  • Complex professional work
  • Everyday knowledge work
  • Cost-aware agentic coding
  • Agentic coding
  • Professional knowledge work
  • Software engineering
  • Professional deliverables
  • Research with computer use
  • Scientific reasoning
Published benchmarks
Terminal-Bench 2.0pass_rate——
  • 82.7%effort xhigh(setup not fully disclosed)
—
Terminal-Bench 2.1pass_rate
  • 88.8%effort max(setup not fully disclosed)
  • 87.4%effort max(setup not fully disclosed)
——
Terminal-Bench 4.0pass_rate———
  • 57.9%(setup not fully disclosed)
BrowseComp (with tools)accuracysetups differ by: reasoning
  • 90.4%effort max(setup not fully disclosed)
  • 87.5%effort max(setup not fully disclosed)
  • 84.4%effort xhigh(setup not fully disclosed)
  • 91.5%(setup not fully disclosed)
Agents' Last Examscoresetups differ by: reasoning
  • 52.7%effort max(setup not fully disclosed)
  • 50.4%effort max(setup not fully disclosed)
—
  • 59.3%(setup not fully disclosed)
DeepSWE 1.1resolved_rate———
  • 74.1%(setup not fully disclosed)
SWE-Bench Pro (Public)pass_ratesetups differ by: reasoning
  • 64.6%effort max(setup not fully disclosed)
  • 63.4%effort max(setup not fully disclosed)
  • 58.6%effort xhigh(setup not fully disclosed)
—
OSWorld 2.0success_rate
  • 62.6%effort max(setup not fully disclosed)
  • 50.2%effort max(setup not fully disclosed)
——
OSWorld 2.0 (v2026.08.08 offline, partial score)score———
  • 72.6%(setup not fully disclosed)
OSWorld-Verifiedsuccess_rate——
  • 78.7%effort xhigh(setup not fully disclosed)
—
GDPval (wins or ties)wins_or_ties——
  • 84.9%effort xhigh(setup not fully disclosed)
—
OpenAI MRCR v2 8-needle 512K–1Maccuracysetups differ by: reasoning
  • 73.8%effort max(setup not fully disclosed)
  • 72.5%effort max(setup not fully disclosed)
  • 74%effort xhigh(setup not fully disclosed)
  • 96.3%(setup not fully disclosed)
FrontierMath Tier 4 (v2)score———
  • 97.6%(setup not fully disclosed)
MMMU Pro (with tools)accuracysetups differ by: reasoning
  • 84.6%effort max(setup not fully disclosed)
  • 82%effort max(setup not fully disclosed)
  • 83.2%effort xhigh(setup not fully disclosed)
—
Humanity's Last Exam Full (with tools)accuracy——
  • 52.2%effort xhigh(setup not fully disclosed)
—
GPQA Diamondaccuracysetups differ by: reasoning
  • 94.6%effort max(setup not fully disclosed)
  • 92.9%effort max(setup not fully disclosed)
  • 93.6%effort xhigh(setup not fully disclosed)
  • 96%(setup not fully disclosed)
Toolathlonscoresetups differ by: reasoning
  • 58%effort max(setup not fully disclosed)
  • 53.1%effort max(setup not fully disclosed)
  • 55.6%effort xhigh(setup not fully disclosed)
—