Google Gemini

Use Gemini for root cause analysis — free tier, 1M context window

Why Gemini?

• Free tier: 1,500 requests/day on gemini-2.0-flash — sufficient for most teams
• 1M context window: Feed the full health history of all servers at once
• No credit card required to get started

Setup

1. Go to aistudio.google.com/app/apikey
2. Click Create API key — free, no credit card needed
3. Set the environment variable:

export GEMINI_API_KEY=AIza...

4. Configure in .langsight.yaml:

investigate:
  provider: gemini
  model: gemini-2.0-flash    # default

Models

Model	Context	Free tier	Best for
gemini-2.0-flash	1M	1,500/day	Default — fast, capable
gemini-2.5-pro	1M	50/day	Best quality analysis
gemini-1.5-flash	1M	1,500/day	Budget alternative

Free tier limits

Model	Requests/day	Requests/minute
gemini-2.0-flash	1,500	15
gemini-2.5-pro	50	2
gemini-1.5-flash	1,500	15

Pricing (after free tier)

Model	Input	Output
gemini-2.0-flash	$0.10/1M	$0.40/1M
gemini-2.5-pro	$1.25/1M	$10/1M
gemini-1.5-flash	$0.075/1M	$0.30/1M

How it works

LangSight uses Gemini via its OpenAI-compatible API endpoint — no extra SDK needed. The openai package (already installed) is pointed at Google’s servers:

POST https://generativelanguage.googleapis.com/v1beta/openai/chat/completions