Skip to content

docs(providers): add the Apodex provider page - #906

Open
zhanghanduo wants to merge 6 commits into
BerriAI:mainfrom
zhanghanduo:litellm_apodex_provider_docs
Open

docs(providers): add the Apodex provider page#906
zhanghanduo wants to merge 6 commits into
BerriAI:mainfrom
zhanghanduo:litellm_apodex_provider_docs

Conversation

@zhanghanduo

@zhanghanduo zhanghanduo commented Aug 16, 2026

Copy link
Copy Markdown

Adds the Apodex provider page for BerriAI/litellm#37071. The main PR already links its README and endpoint capability table to /docs/providers/apodex, so this page should merge before or with it

The page covers APODEX_API_KEY, the optional APODEX_API_BASE, supported models, pricing, Chat Completions, Responses, Anthropic Messages compatibility, streaming, tools, and model-specific parameter support

It distinguishes the two provider contracts: core models accept native sampling and function tools, while Deep Research tiers use a server-side agent and external tools through mcp_servers. Core Messages requests use Apodex's native endpoint, while Deep Research Messages requests are translated through the Responses API

The Responses section documents stateless core-model restrictions and notes that LiteLLM validates top-level parameters. Values passed through extra_body remain a raw provider escape hatch and bypass that validation

@vercel

vercel Bot commented Aug 16, 2026

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

Project Deployment Actions Updated (UTC)
litellm Ready Ready Preview Aug 16, 2026 10:25am

Request Review

@zhanghanduo

Copy link
Copy Markdown
Author

Please review this before BerriAI/litellm#37071. It documents both required environment variables, and the Vercel preview is green

Covers both model families, the parameter contract each one accepts on
/chat/completions and /responses, native Anthropic Messages for the core
models, and cost tracking with cached input.
- the core models default stream to false; only the Deep Research tiers
  default it to true
- the Deep Discover tier is served on /responses only
- state the 65,536 max output GET /v1/models reports for the core models
- scope the 200K multiplier to the core models and to the three price
  components it applies to
- name what falls outside cost tracking: the per-invocation Deep Research
  tool call billing, and the unregistered 1.0 tiers
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant