wehub-resource-sync
15 lines
827 B
TOML
15 lines
827 B
TOML
name = "Baseten"
|
|
npm = "@ai-sdk/openai-compatible"
|
|
# Raw HTTP reasoning controls (sources accessed 2026-06-25):
|
|
# POST https://inference.baseten.co/v1/chat/completions uses top-level
|
|
# chat_template_args.enable_thinking = true for documented opt-in models and
|
|
# top-level reasoning_effort for DeepSeek V4 Pro and GPT OSS 120B. No explicit
|
|
# false toggle behavior or reasoning-token budget field/bounds are documented.
|
|
# The strict OpenAPI omits reasoning_effort but allows chat_template_args;
|
|
# model support is established by the model-specific reasoning guide, not schema.
|
|
# https://docs.baseten.co/inference/model-apis/reasoning
|
|
# https://docs.baseten.co/reference/inference-api/chat-completions
|
|
doc = "https://docs.baseten.co/inference/model-apis/overview"
|
|
api = "https://inference.baseten.co/v1"
|
|
env = ["BASETEN_API_KEY"]
|