From f7632ab931bec58f23036eb9a1a5b96d91ae391f Mon Sep 17 00:00:00 2001 From: tin Date: Thu, 6 Aug 2026 04:59:40 +0000 Subject: [PATCH] docs(router): clarify thinking budget warning Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com> --- litellm/router_strategy/complexity_router/README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/litellm/router_strategy/complexity_router/README.md b/litellm/router_strategy/complexity_router/README.md index 6c202d4dcc5..b5c78cb8bed 100644 --- a/litellm/router_strategy/complexity_router/README.md +++ b/litellm/router_strategy/complexity_router/README.md @@ -109,8 +109,8 @@ take precedence over alias `litellm_params` defaults. Parameters unsupported by the selected model follow LiteLLM's existing `drop_params` behavior. Unknown parameter names produce a configuration warning but remain available for provider-specific parameters. `thinking` is the supported structured parameter; -`thinking_budget` is not a top-level LiteLLM parameter and therefore produces a -warning, as intended. +`thinking_budget` is not a top-level LiteLLM parameter, so a tier naming it gets +a warning telling it to use `thinking: {type: enabled, budget_tokens: N}`. ### Full Configuration