Google Gemini routing

Use Gemini and Vertex AI from one endpoint

Run TEKIZ.AI on a machine or server you control with your Google AI Studio keys or Vertex AI project. Your tools use one router address for Google models and other services you approve. Hosted access uses a separate TEKIZ.AI key.

Set up a local router Create hosted key Read quickstart Browse models Integrations

Use the interface your client expects

On a router you control, use Google and Vertex models through an OpenAI-compatible client. Gemini model discovery and generateContent are available where supported.

Tool calls stay structured

TEKIZ.AI translates OpenAI-style tools for Gemini and returns Gemini function calls as normal OpenAI-compatible tool_calls.

When a Gemini or Vertex route rate-limits

If a Gemini key, Vertex project, or model is rate-limited, TEKIZ.AI can try another authorized route you configured.

Send a test request

export OPENAI_BASE_URL=https://api.tekiz.ai/v1
export OPENAI_API_KEY=sk_tekiz_your_generated_key

curl "$OPENAI_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "tekizai/agentic",
    "messages": [{"role": "user", "content": "Explain when a lookup tool is useful."}],
    "tools": [{
      "type": "function",
      "function": {
        "name": "lookup",
        "parameters": {"type": "object", "properties": {"query": {"type": "string"}}}
      }
    }]
  }'

Send files to services that accept them

Images, audio, video, and documents are sent only to models that can handle that kind of input.

Use more than one account key or project

TEKIZ.AI can share requests across the Google keys and projects you connect, then pause only the one that is unavailable.

Use accounts you are allowed to use

TEKIZ.AI does not grant Google access. Connect a key or project you are authorized to use, or use a hosted TEKIZ.AI plan.

Why put TEKIZ.AI in front of Gemini?

Your tools keep one address while TEKIZ.AI checks keys, keeps tool calls in the expected format, sends files to compatible models, and tries an available backup when Gemini cannot take the request.

Next steps

Create an API key, copy the hosted quickstart, browse Gemini-capable routes, or request managed Gemini review.

Stay within your approved services

If your company approves only Google AI, keep that boundary. TEKIZ.AI can reduce token burn by choosing eligible models. Your Google AI or Vertex AI project handles provider billing.