Gemini 3.5 Flash-Lite on Vertex: does maxOutputTokens bound response + thinking?

Gemini API Documentation
1

For the managed gemini-3.5-flash-lite publisher model on Vertex/Agent Platform v1 generateContent (not the Developer API, Live API or a tuned endpoint), does generationConfig.maxOutputTokens = G with thinkingConfig.thinkingLevel = MINIMAL enforce usageMetadata.thoughtsTokenCount + usageMetadata.candidatesTokenCount <= G for a single-turn, non-streaming text/JSON response?

Input is a silent video plus text, with no tools, grounding or explicit cache.

Please identify the applicable official specification. If reasoning is outside that limit, what provider-enforced maximum covers all chargeable reasoning tokens? Please also clarify how this bound applies if the client times out but generation finishes server-side.

We have not enabled inference to test this assumption.

2

Hello @MintGun ,

For issues related to the Vertex API, please contact your sales representative if you have one. Otherwise, you can find more relevant assistance on the Google Cloud Forum.