What is context length, and what limits apply to it?

Context length is the combined number of input and output tokens in a request. Two main limits can apply:

• Model context window

  • Each model has a maximum context window. If the combined input and requested output exceeds this limit, the request may be rejected with an error such as: input is too long (received … tokens; max …).

  • Check the v1/models endpoint to find the exact context length supported by each model.

• Per-tier output cap

  • Your usage tier may impose a separate limit on the number of output tokens. If the response reaches this limit, it may be truncated with: finish_reason: "length".

  • Upgrade your tier to access a higher output limit. See our docs for more details on Tiers.