Day 0 support: Gemini 3.7 Flash

LiteLLM now supports gemini-3.7-flash on day 0, on both Google AI Studio (gemini/) and Vertex AI (vertex_ai/). Google's newest Flash model delivers faster responses with meaningfully better quality than 3.6 Flash and scores higher on complex multi-step agentic, coding, and reasoning benchmarks.

note

No Docker image upgrade needed. Gemini 3.7 Flash routes through the existing Gemini configs, so any recent LiteLLM version works out of the box for inference. For cost tracking, hit the Reload Model Cost Map button in the Admin UI (or POST /reload/model_cost_map) to pull the latest pricing from GitHub. This is available on v1.76.0 and above. The gemini-3.7-flash pricing and metadata are also bundled starting in v1.98.0-dev.2 for anyone running with LITELLM_LOCAL_MODEL_COST_MAP=true.

Launch pricing

Gemini 3.7 Flash launches at a 50% discount that runs through December 31, 2027. LiteLLM tracks cost at the promotional rate.

	Promotional	Standard
Input	$0.75 / 1M tokens	$1.50 / 1M tokens
Output	$3.75 / 1M tokens	$7.50 / 1M tokens

Cache reads, batch, flex, and priority tiers are discounted proportionally.

Quick Start

SDK

from litellm import completion

response = completion(
    model="gemini/gemini-3.7-flash",
    messages=[{"role": "user", "content": "Summarize this article in 3 bullet points."}],
)

print(response.choices[0].message.content)

PROXY

1. Setup config.yaml

model_list:
  - model_name: gemini-3.7-flash
    litellm_params:
      model: gemini/gemini-3.7-flash
      api_key: os.environ/GEMINI_API_KEY

  # Or use Vertex AI
  - model_name: vertex-gemini-3.7-flash
    litellm_params:
      model: vertex_ai/gemini-3.7-flash
      vertex_project: your-project-id
      vertex_location: us-central1

2. Start proxy

litellm --config /path/to/config.yaml

3. Make requests

curl -X POST http://localhost:4000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <YOUR-LITELLM-KEY>" \
  -d '{
    "model": "gemini-3.7-flash",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Thinking levels

Gemini 3.7 Flash is a reasoning model. LiteLLM maps OpenAI reasoning_effort to Gemini's thinkingLevel, so the same request shape you use for other reasoning models works here.

from litellm import completion

response = completion(
    model="gemini/gemini-3.7-flash",
    messages=[{"role": "user", "content": "What's 2+2?"}],
    reasoning_effort="low",
)

print(response.choices[0].message.content)

Known limitation at launch

The minimal thinking level is not yet supported on gemini-3.7-flash. The Gemini API returns a 400 (Thinking level MINIMAL is not supported for this model). Google plans minimal thinking support as a fast follow. All other thinking levels work as expected.

Supported Endpoints

LiteLLM provides full end-to-end support for Gemini 3.7 Flash on:

• /v1/chat/completions - OpenAI-compatible chat completions endpoint
• /v1/responses - OpenAI Responses API endpoint (streaming and non-streaming)
• /v1/messages - Anthropic-compatible messages endpoint
• /v1/generateContent - Google Gemini API compatible endpoint

All endpoints support streaming and non-streaming responses, function calling with thought signatures, multi-turn conversations, and full multimodal input (text, image, audio, video).

Feedback

Running Gemini 3.7 Flash through LiteLLM and hitting something unexpected? Share it on GitHub discussion #36799.