Google Gemini models

Choose the right model in the family

Start with the smallest model likely to meet the workload. Compare it with one higher-capability candidate on the same tasks, then choose by accepted-result quality, latency, reliability, and total cost.

• Start with the stable Gemini 3.7 Flash release for multimodal coding and AI agent workflows.
• Keep Gemini 3.6 Flash when its established behavior matters more than the newest features.
• Use preview models only when they can be switched off quickly and a tested fallback is ready.

Inputs, outputs, context and tools

• Gemini 3.7 Flash accepts text, image, video, audio and PDF input and returns text.
• It publishes a 1,048,576-token input limit and 65,536-token output limit.
• Supported features include function calling, structured output, code execution, search grounding, file search and URL context.
• Thinking levels are low, medium and high; medium is the documented default.

Where this family fits

These are strong starting points—not universal recommendations. Test them with your own prompts, tools and acceptance criteria.

• Understanding video, audio, and documents through one request format.
• Coding agents that combine structured tools with large context windows.
• Analysis backed by web search when current facts matter.
• Large-scale extraction and classification with explicit schemas.

Reasoning, context and cost controls

Do not compare only the headline input rate. Include output, cache reads and writes, long-context tiers, reasoning tokens, tool calls, retries, and the amount of history resent on every turn.

• Use context caching for repeated large prefixes and compare the cache economics with real traffic.
• Batch processing can reduce vendor cost for asynchronous workloads, but route support must be confirmed.
• Pin an exact stable ID for production instead of relying on a moving preview alias.

Send a minimal request

This example uses the first provider model ID in the table. If Model Plaza shows a different ID for your API key, copy the Model Plaza value exactly.

curl https://app.llmfly.ai/v1/chat/completions \
  -H "Authorization: Bearer $LLMFLY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3.1-pro-preview",
    "messages": [{"role": "user", "content": "Explain this request in three bullets."}],
    "stream": false
  }'

Limits and compatibility notes

• Gemini 3.7 Flash does not publish audio or image generation and does not support the Live API.
• Computer use is preview functionality and needs separate risk controls.
• Pricing and preview status are date-sensitive; recheck them before launch.

Production evaluation checklist

• Run 20–50 representative tasks and define pass criteria before testing.
• Measure time to first token, total latency, input and output tokens, tool retries, and cost per successful task.
• Test the endpoint, streaming mode, tool schemas, and structured output used by the application.
• Verify 400, 401, 404, 429, and temporary 5xx handling with bounded retries.
• Keep the model ID in configuration and validate a fallback before moving production requests.

What to confirm in Model Plaza

Model Plaza shows which model IDs your API key can use, their billing multipliers, availability, supported endpoints, and compatibility notes. Confirm these details with a small request before deployment.

• Gemini 3.7 Flash supports low, medium and high thinking levels.
• Its published input modalities include text, image, video, file and audio.
• Preview aliases can change; prefer a stable exact ID for production when one meets the task.

Move from model research to a usable model ID

Open a model page above, note its provider model ID, and then use Model Plaza to confirm the ID available to your API key. Store that ID in configuration so it can be reviewed and changed without rewriting the application.