Gemini 2.0 updates

[Dual-0]

Google has expanded the Gemini 2.0 model family, focusing on improved performance and cost-effectiveness. Gemini 2.0 Flash (generally available), Gemini 2.0 Flash-Lite (Public Preview), and an experimental Gemini 2.0 Pro are now available.

  - gemini-2.0-flash
  - gemini-2.0-flash-lite-preview-02-05
  - gemini-2.0-pro-exp-02-05
  - gemini-2.0-flash-thinking-exp-01-21

Models in API Docs

edit: fixed in #5703

[ditcho]

You can just add these to your .env and set your location to us-central1 to access these via vertex

[Dual-0]

When I used it via Gemini API (AI Studio) the new gemini-2.0-flash-lite-preview-02-05 and gemini-2.0-pro-exp-02-05 doesnt work.

2025-02-06 12:27:30 error: [GoogleClient] There was an issue generating the completion [GoogleGenerativeAI Error]: Error fetching from https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash-lite-preview-02-05:streamGenerateContent?alt=sse: [400 Bad Request] HARM_CATEGORY_CIVIC_INTEGRITY threshold cannot be 5

[ditcho]

That might have something to do with these settings in the .env:

# BLOCK_NONE is available by default, no special account requirements.
#
# Available options: BLOCK_NONE, BLOCK_ONLY_HIGH, BLOCK_MEDIUM_AND_ABOVE, BLOCK_LOW_AND_ABOVE
#
# GOOGLE_SAFETY_SEXUALLY_EXPLICIT=BLOCK_ONLY_HIGH
# GOOGLE_SAFETY_HATE_SPEECH=BLOCK_ONLY_HIGH
# GOOGLE_SAFETY_HARASSMENT=BLOCK_ONLY_HIGH
# GOOGLE_SAFETY_DANGEROUS_CONTENT=BLOCK_ONLY_HIGH
# GOOGLE_SAFETY_CIVIC_INTEGRITY=BLOCK_ONLY_HIGH```

[Dual-0]

GOOGLE_SAFETY_CIVIC_INTEGRITY=BLOCK_NONE doesnt work.
With GOOGLE_SAFETY_CIVIC_INTEGRITY=BLOCK_ONLY_HIGH it works... weird.

thanks @ditcho

[Dual-0]

FYI I can not use gemini-2.0-flash-lite-preview-02-05 for titling

2025-02-07 09:37:07 warn: Current vision model does not support titling without an attachment; falling back to default model gemini-1.5-flash-latest

edit: fixed in #5703