$300 in Free Google Credits for Media and Coding

Google Cloud gives every new account $300 in free credits, valid for 90 days. Most people use this to spin up a VM, run a tutorial, and forget about it. That's a waste. Those credits unlock access to over 200 models in Vertex AI Model Garden, including Google's own Gemini, Imagen, Veo, Chirp, and Lyria, plus third-party models like Claude and open-source options like Llama and Gemma. You also get Gemini CLI, Google's open-source terminal coding agent.

Link: cloud.google.com

What $300 unlocks

Your credits cover every Google Cloud service. For AI, the interesting parts live in Vertex AI Model Garden and Gemini CLI.

Google first-party models

Gemini (LLMs): Gemini 3.1 Pro with 1M token context for complex reasoning. Gemini 3 Flash for fast agentic coding. Gemini 3.1 Flash-Lite for cheap high-volume calls. Plus the 2.5 generation (Pro, Flash, Flash-Lite) and a Live API for real-time bidirectional audio streaming.

Imagen 4 (Image Generation): Three tiers. Fast for cheap iteration and thumbnails. Standard for balanced quality. Ultra for highest fidelity photorealistic output. Up to 4K resolution, multi-reference composition with up to 14 images, aspect ratio control.

Veo 3.1 (Video): Text-to-video, image-to-video, first/last frame control for transitions, video extension for chaining clips. Audio generation built in. Three tiers: Standard (best quality), Fast (mid), Lite (50% cheaper than Fast).

Chirp 3 HD (TTS): Text-to-speech across 35 languages. Custom voice cloning from ~10 seconds of audio (gated behind diligence approval).

Lyria (Music): Lyria 2 (GA) for full tracks from text. Lyria 3 Clip (preview) for 30-second clips. Lyria 3 Pro (preview) for tracks from text + image prompts.

Third-party and open-source

Claude (Anthropic), Llama 3.2, Mistral, Gemma 4, and dozens more through Model Garden. Same API surface, same credits. One billing account for everything.

Gemini CLI

Google's open-source terminal coding agent. Uses Gemini with a ReAct loop, built-in tools for file ops, shell commands, Google Search grounding, and MCP server support. Free tier gives you 1,000 requests/day with a personal Google account. With your GCP credits, you get higher limits through Vertex AI.

Repo: github.com/google-gemini/gemini-cli

Gemini CLI setup

Gemini CLI is the fastest way to start burning your credits on something useful. It's the Google equivalent of Claude Code.

Install:

npm install -g @google/gemini-cli

Or try it without installing:

npx @google/gemini-cli

To route through your GCP project (using your $300 credits instead of the rate-limited free tier):

export GOOGLE_CLOUD_PROJECT=your-project-id
gcloud auth application-default login

Now you get Gemini 3.1 Pro with a 1M token context window. Built-in tools handle file read/write, shell commands, search grounding, and web fetching. MCP support means you can plug in the same servers you already use with Claude Code.

The tradeoff: Gemini's coding output is less consistent than Claude's for complex refactors. For straightforward tasks (file generation, test writing, simple debugging), it's solid. Best used as a complement when you've hit your Claude Code limits for the day.

Imagen 4: image generation setup

You need the Vertex AI API enabled and a service account. This setup applies to all media models (Imagen, Veo, Chirp, Lyria).

Open Google Cloud Console, select your project
Go to APIs & Services > Enable APIs, search "Vertex AI API", click Enable

Create a service account:

gcloud iam service-accounts create media-gen \
  --display-name="Media Generation"

gcloud projects add-iam-policy-binding YOUR_PROJECT_ID \
  --member="serviceAccount:media-gen@YOUR_PROJECT_ID.iam.gserviceaccount.com" \
  --role="roles/aiplatform.user"

gcloud iam service-accounts keys create ~/gcp-media-key.json \
  --iam-account=media-gen@YOUR_PROJECT_ID.iam.gserviceaccount.com

Install the gemini-media-mcp server to control all media models from Claude Code:

uvx gemini-media-mcp setup

Or add it manually to your MCP config:

{
  "mcpServers": {
    "gemini-media": {
      "command": "uvx",
      "args": ["gemini-media-mcp"],
      "env": {
        "GOOGLE_GENAI_USE_VERTEXAI": "true",
        "GOOGLE_CLOUD_PROJECT": "your-project-id",
        "GOOGLE_CLOUD_LOCATION": "us-central1",
        "GOOGLE_APPLICATION_CREDENTIALS": "/path/to/gcp-media-key.json"
      }
    }
  }
}

Repo: github.com/CxOAGI/gemini-media-mcp