Gemini Provider

Google Gemini is Google’s multimodal AI model family. Gemini offers a generous free tier, making it an excellent choice for getting started without any cost.

Get an API Key

Go to Google AI Studio
Sign in with your Google account
Click Get API Key in the top navigation
Click Create API key and select a Google Cloud project (one is created automatically)
Copy the key

Your key will look like: AIza...

Configure VibeCody

Option 1: Environment variable (recommended)

export GEMINI_API_KEY="AIza..."
vibecli --provider gemini

Option 2: Config file (~/.vibecli/config.toml)

[gemini]
enabled = true
api_key = "AIza..."
model = "gemini-2.5-flash"

Option 3: API key helper

[gemini]
enabled = true
api_key_helper = "~/.vibecli/get-key.sh gemini"
model = "gemini-2.5-pro"

Model Selection

Model	Strengths	Context	Best for
`gemini-2.5-pro`	Highest quality, deep reasoning	1M tokens	Complex coding, architecture, analysis
`gemini-2.5-flash`	Fast with thinking capabilities	1M tokens	Balanced quality and speed
`gemini-2.0-flash`	Very fast, cost-effective	1M tokens	Quick tasks, high-volume use

Default: gemini-2.5-flash

Override from the CLI:

vibecli --provider gemini --model gemini-2.5-pro

Free Tier

Gemini offers one of the most generous free tiers among cloud providers:

Model	Free rate limit	Free context
Gemini 2.0 Flash	15 RPM, 1M TPM	1M tokens
Gemini 2.5 Flash	5 RPM, 250K TPM	1M tokens
Gemini 2.5 Pro	2 RPM, 250K TPM	1M tokens

RPM = requests per minute, TPM = tokens per minute.

The free tier is sufficient for personal coding use. For higher throughput, enable billing in Google Cloud Console.

Paid Pricing

For usage beyond the free tier (check ai.google.dev/pricing for current rates):

Model	Input (per 1M tokens)	Output (per 1M tokens)
Gemini 2.5 Pro	$1.25 - $2.50	$10.00 - $15.00
Gemini 2.5 Flash	$0.15 - $0.30	$0.60 - $3.50
Gemini 2.0 Flash	$0.10	$0.40

Pricing varies by context length (under/over 200K tokens).

Multimodal Capabilities

Gemini natively supports image understanding. In VibeUI, you can:

Paste screenshots of UI mockups and ask Gemini to generate code
Share error screenshots for debugging
Analyze architecture diagrams

This makes Gemini particularly useful for front-end development workflows.

1M Token Context Window

All Gemini models support a 1 million token context window, which means you can:

Load entire codebases into context
Analyze very large files without truncation
Maintain long conversation histories

VibeCody’s infinite context manager works especially well with Gemini’s large context window.

Verify Connection

vibecli --provider gemini -c "Say hello and identify yourself"

Troubleshooting

Invalid API key

Error: 400 - API key not valid

Verify the key at Google AI Studio
Ensure the key starts with AIza
Check that the associated Google Cloud project has the Generative Language API enabled

Rate limited (free tier)

Error: 429 - Resource has been exhausted

You have exceeded the free tier rate limit
Wait 60 seconds for the rate limit to reset
Consider enabling billing for higher limits
Use gemini-2.5-flash for the most generous free-tier limits

Region restrictions

Some regions may not have access to all Gemini models. If you encounter availability issues:

Check Google AI availability for your region
Try a different model (Flash models have broader availability)
Use a VPN if permitted by your organization

API key vs. OAuth

VibeCody uses API key authentication (not OAuth). Ensure you are generating an API key from Google AI Studio, not an OAuth client credential from Google Cloud Console.

Billing not enabled

Error: 403 - Billing account not found

If you exceed free-tier limits, you need to enable billing:

Go to Google Cloud Console
Select the project linked to your API key
Navigate to Billing and link a billing account

Provider: Gemini