Gemini 3.1 Pro API: Model ID, Pricing, Context Window and How to Call It (2026)

Short answer

Gemini 3.1 Pro is the Pro model Google released on February 19, 2026. It is still a preview model, with the model ID gemini-3.1-pro-preview, and TeamoRouter uses the same ID. It takes up to 1,048,576 input tokens and returns up to 65,536. Google's list price is $2 per million input tokens and $12 per million output tokens (output includes thinking tokens), or $4 / $18 when a prompt is over 200K tokens. On the day of writing (October 22, 2026) TeamoRouter charges about 49% of the official price; the live price is on the model page.

What Gemini 3.1 Pro is, and is it still worth using?

Google describes Gemini 3.1 Pro as its third-generation Pro model for multimodal understanding, agentic work and coding. Compared with Gemini 3 Pro, Google says it thinks better, uses tokens more efficiently and stays more factually grounded, and that it is tuned for software engineering and multi-step tasks that depend on precise tool use. The previous gemini-3-pro-preview was shut down on March 9, 2026, and that name now points to 3.1 Pro.

On October 13, 2026, Gemini 4 Argon became generally available in the Gemini API. Official prices per million tokens, prompts up to 200K:

Which to pick:

• New project, or you need very long output: try Argon first. On the day of writing, Argon on TeamoRouter costs about 29% of its list price versus about 49% for 3.1 Pro, so Argon is cheaper.
• Prompts and tool calls already tuned for 3.1 Pro: keep using it. At Google's list prices, 3.1 Pro becomes cheaper than Argon once Argon moves to $4 / $20 on December 1.

As of October 22, 2026, Google's deprecations page lists no shutdown date for 3.1 Pro.

Specs at a glance

From Google's Gemini 3.1 Pro model page:

Google also offers a gemini-3.1-pro-preview-customtools endpoint that gives priority to your own custom tools. TeamoRouter offers gemini-3.1-pro-preview.

Pricing

Google list price (Gemini API pricing, Standard tier, per million tokens):

Cache storage is an extra $4.50 per million tokens per hour. Google's Batch and Flex tiers are half the Standard price.

TeamoRouter bills per token at about 49% of the official price on the day of writing, with no monthly fee. Prices follow upstream changes; current numbers are on the model page and the pricing page.

Thinking tokens are billed as output. The default high level thinks the most, so each request costs the most; for simple tasks, low is faster and cheaper.

Call it through TeamoRouter

Google's list of regions where the Gemini API is available leaves out mainland China, Hong Kong and Russia, among others. TeamoRouter offers both the OpenAI-compatible format and the native Gemini format, so existing code only needs a new base URL and key. Create a key in the dashboard.

OpenAI-compatible format (base URL includes /v1):

from openai import OpenAI

client = OpenAI(base_url="https://api.teamorouter.vip/v1", api_key="YOUR_TEAMOROUTER_KEY")
resp = client.chat.completions.create(
    model="gemini-3.1-pro-preview",
    reasoning_effort="low",
    messages=[{"role": "user", "content": "Find the race condition in this code: ..."}],
)
print(resp.choices[0].message.content)

reasoning_effort maps to Gemini's thinking level (low, medium, high); leave it out to get Google's default, high.

Native Gemini format (no /v1 in front of the path):

curl https://api.teamorouter.vip/v1beta/models/gemini-3.1-pro-preview:generateContent \
  -H "Authorization: Bearer YOUR_TEAMOROUTER_KEY" \
  -H "content-type: application/json" \
  -d '{
    "contents": [
      {"role": "user", "parts": [{"text": "Explain a context window in one sentence."}]}
    ]
  }'

Google recommends leaving temperature at its default of 1.0 for Gemini 3 models; lower values can cause looping or worse results. Full reference: API integration.

Use it in Gemini CLI and Cherry Studio

Gemini CLI: set three environment variables, then pick the model with -m:

export GOOGLE_GEMINI_BASE_URL="https://api.teamorouter.vip"
export GEMINI_API_KEY="YOUR_TEAMOROUTER_KEY"
export GEMINI_API_KEY_AUTH_MECHANISM="bearer"
gemini -m gemini-3.1-pro-preview

Install steps: Gemini CLI setup.

Cherry Studio: Settings → Model Services → Add, choose Gemini as the provider type, set the API URL to https://api.teamorouter.vip, fetch the model list and select gemini-3.1-pro-preview. See Cherry Studio setup.

Paying for usage

No subscription: you top up a balance from $5 and pay only for the tokens you use, by card or Alipay. The same balance covers Claude, GPT and the other Gemini models.

FAQ

Will gemini-3.1-pro-preview be shut down? As of October 22, 2026, Google has not announced a date. It is a preview model, and the previous preview, gemini-3-pro-preview, was shut down on March 9, so long-running projects should watch Google's deprecations page.

Can I turn thinking off? Not completely. The lowest level for 3.1 Pro is low; minimal is not supported.

How are prompts over 200K tokens billed? Google's list price is $4 input and $18 output. The live TeamoRouter price is on the model page.

Can I use it in Codex? No. Codex sends requests through the Responses API, and TeamoRouter's /v1/responses supports GPT models only; Gemini requests return 400. For Gemini in the terminal, use Gemini CLI.

Do I need a Google account? No. One TeamoRouter key covers Gemini, Claude and GPT.

Next steps

• Gemini 3.1 Pro model page: live price.
• Create an API key and send your first gemini-3.1-pro-preview request.
• Gemini 4 Argon API: Google's newest flagship.
• Gemini 3.8 Flash API: the cheaper, faster Flash model.