Skip to content

Graphorin API reference v0.13.8


Graphorin API reference / @graphorin/provider / / TokenLimitParam

Type Alias: TokenLimitParam

ts
type TokenLimitParam = "max_tokens" | "max_completion_tokens";

Defined in: packages/provider/src/internal/openai-shaped.ts:52

Stable

Wire name for the completion-token ceiling. The long tail of OpenAI-compatible servers (llama.cpp, LM Studio, vLLM, LocalAI) expects max_tokens; current OpenAI models (GPT-5 family, o-series) reject it with HTTP 400 and require max_completion_tokens.