Google Gemini

Use Google Gemini models as the LLM for your agents.

Get an API key

Get a free key at aistudio.google.com under Get API key.

Setup

pai create model-provider google \
  --provider gemini \
  --api-key AIzaSy...

This stores your API key securely and creates a ModelProvider that covers every Gemini model under your subscription.

Verify:

pai get model-providers
# NAME     PROVIDER   ENDPOINT   MAX/DAY   LAST USED   AGE
# google   gemini                          —           5s

Supported models

Model	Reference	Best for
Gemini 2.5 Flash	google/gemini-2.5-flash	Best overall — fast, high token limits, cheap
Gemini 2.0 Flash	google/gemini-2.0-flash	Stable previous generation
Gemini 2.5 Pro	google/gemini-2.5-pro	Deep reasoning, complex tasks

Use in an agent

spec:
  models:
    - google/gemini-2.5-flash

Multiple models — first is primary, rest are fallbacks:

spec:
  models:
    - google/gemini-2.5-flash     # primary
    - google/gemini-2.5-pro       # fallback for hard questions

The Pai Gateway translates Anthropic-style /v1/messages requests to Gemini's format and back, so harness-backed agents using Claude Code / the Anthropic SDK work against Gemini without code changes.

Token budgets

Cap how many tokens this subscription burns per day across every agent — a hard safety net against runaway spend.

apiVersion: pai.io/v1
kind: ModelProvider
metadata:
  name: google
spec:
  provider: gemini
  apiKeySecretRef:
    name: google-key
    key: api-key
  maxTokensPerDay: 5000000      # daily cap shared across all agents
  maxTokensPerRequest: 1000000  # Gemini supports a 1M-token context window

When the daily cap is hit, the gateway returns HTTP 429 until midnight UTC. Agents that list another provider in spec.models automatically fall over to it.

Expose via the LLM Gateway

Set externalAccess.enabled: true to let developers outside the cluster — laptops, CI, scripts — route their own LLM traffic through this provider. The Google API key stays inside Pai; clients authenticate with a Pai AccessKey instead.

spec:
  externalAccess:
    enabled: true
    maxTokensPerDay: 2000000    # separate budget for external usage

Once enabled, developers connect with three commands:

pai login https://api.pairun.dev --access-key pak_...
eval $(pai gateway env)
claude   # Claude Code routes through Pai → Gemini via on-the-fly translation

See LLM Gateway for the full onboarding flow, AccessKey management, and per-developer rate limits.

Access control

Narrow which models agents may call on this provider with allowedModels / deniedModels, or attach prompt-injection guards. See Security controls on the Model page for the full field list.