Skip to main content

GPT 5.5

OpenAI's flagship reasoning chat model — long-context, vision-capable, streamed, with the full range of reasoning effort from instant replies to X-High deliberation.

Capabilities​

FeatureSupport
Chat (text-to-text)Yes, streaming
Vision (image understanding)Yes — up to 10 connected images
Document referencesYes — up to 100 text/document inputs
Context Window1,050,000 tokens
Max Output128,000 tokens
Reasoning EffortNone / Low / Medium / High / X-High

Chat Nodes on the Canvas​

Same wiring as the Claude models: add an AI Chat node, connect images and text upstream as context, and iterate in the node's conversation. Downstream nodes can consume the reply — e.g. feed a generated shot list into a GPT Image 2 Agent node, which uses GPT 5.5 internally as its planner.

The Widest Effort Range​

GPT 5.5 spans None through X-High in one model:

  • None — instant conversational replies, no reasoning pass.
  • Low / Medium — everyday writing and analysis.
  • High / X-High — deep planning, complex constraint problems, meticulous long-document review.

That makes it a good single-model default if you'd rather move a slider than switch models.

What It's Best At​

  • Structured planning — shot lists, production breakdowns, prompt engineering for the image/video models.
  • Instruction-dense tasks — long briefs with many explicit constraints.
  • Second opinions — cross-check a Claude draft's logic (different model family, different blind spots).

Limitations​

  • Text out only — chain to image/video nodes for rendering
  • No audio understanding

See Also​