How do I use Gemini in OpenClaw?
OpenClaw runs Gemini through its built-in google provider with a GEMINI_API_KEY, or through a standardcompute provider in ~/.openclaw/openclaw.json (baseUrl https://api.stdcmpt.com/v1) with standardcompute/google/gemini-3.8-flash as the primary model. The second bills a plan from $19 a month that returns HTTP 402 to new requests once the compute budget is used.
{
"models": {
"mode": "merge",
"providers": {
"standardcompute": {
"baseUrl": "https://api.stdcmpt.com/v1",
"apiKey": "${STANDARDCOMPUTE_API_KEY}",
"api": "openai-responses",
"models": [
{
"id": "google/gemini-3.8-flash",
"name": "Gemini 3.8 Flash",
"reasoning": true,
"input": ["text", "image"],
"contextWindow": 400000,
"maxTokens": 32000
},
{
"id": "google/gemini-3.1-pro-preview",
"name": "Gemini 3.1 Pro Preview",
"reasoning": true,
"input": ["text", "image"],
"contextWindow": 400000,
"maxTokens": 32000
}
]
}
}
},
"agents": {
"defaults": {
"model": { "primary": "standardcompute/google/gemini-3.8-flash" }
}
}
}
"mode": "merge" keeps OpenClaw's own google provider and any key you set for it; only the standardcompute/... refs use this endpoint.
Gemini sign-in is no longer offered
OpenClaw's Google docs say it no longer sets up Gemini CLI or Antigravity sign-in: Google ended consumer Gemini CLI sign-in on June 18, 2026, and Antigravity's terms bar third-party tools. Existing Gemini CLI profiles still run, but OpenClaw cannot create or repair them. The supported routes are an AI Studio key, Vertex AI, or a gateway key.
What an OpenClaw day on Gemini costs
| Per day | 3.8 Flash, 2026 | 3.8 Flash, 2027 | 3.1 Pro |
|---|---|---|---|
| Typical day | $1.54 | $3.08 | $4.19 |
| Busy day | $16 | $31 | $43 |
Gemini API paid tier per million tokens: 3.8 Flash $0.75 input, $0.075 cached, $3.75 output through December 31, 2026, then $1.50, $0.15, $7.50; 3.1 Pro $2, $0.20, $12 up to 200K. Source. The Agent Burn Index measures coding agents such as OpenCode, Cline and Kilo; OpenClaw is not broken out, so these figures use its median and 90th-percentile coding-agent day (September 13 to October 12, 2026) as a stand-in. Priced at the list rates above; your own days will differ.
What works with Gemini in OpenClaw
- Both models are marked text and image, so attachments reach Gemini directly and no imageModel is needed.
- /think levels pass through as reasoning effort; Gemini always reasons, so /think off leaves it at its default.
/model standardcompute/google/gemini-3.1-pro-previewmoves one session to 3.1 Pro.
The catch
OpenClaw's google provider (openclaw onboard --auth-choice gemini-api-key) pays Google directly and refreshes its Gemini model list from Google itself. On either route, the 30-minute heartbeat is part of the bill; the OpenClaw DeepSeek guide lists the settings that shrink it.
Where Standard Compute fits
Gemini 3.8 Flash as primary with Claude Sonnet 5.5 in agents.defaults.model.fallbacks, both on one key, under one monthly cap.
FAQ
Which Gemini model should OpenClaw use?
3.8 Flash as the primary for everyday turns and 3.1 Pro for agents that act on untrusted input, since OpenClaw's security docs recommend the strongest tier for tool-enabled bots.