> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ai-stats.phaseo.app/llms.txt
> Use this file to discover all available pages before exploring further.
# Start Here
> Get your first successful AI Stats API request in under 5 minutes, then choose the right path for SDKs, routing, agents, or production rollout.
Use this page when you want to get moving quickly and do not yet know which part of the docs you need.
You can think of AI Stats in one sentence:
* one API for models, providers, routing, pricing, and async AI workflows
## Fastest path
If you only want the shortest route to first success:
1. Create an API key in the dashboard.
2. Send one request to a `:free` model.
3. Confirm the response shape.
4. Then choose the integration path that matches your app.
Free models can be used with zero deposited credits. Paid models require available wallet balance.
## First request
```bash cURL theme={null}
curl https://api.phaseo.app/v1/responses \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "google/gemma-3-27b:free",
"input": "Reply with: docs ready",
"temperature": 0
}'
```
```typescript TypeScript SDK theme={null}
import AIStats from "@ai-stats/sdk";
const client = new AIStats({ apiKey: process.env.AI_STATS_API_KEY! });
const response = await client.generateResponse({
model: "google/gemma-3-27b:free",
input: "Reply with: docs ready",
temperature: 0,
});
console.log(response.output_text);
```
```python Python SDK theme={null}
from ai_stats import AIStats
client = AIStats(api_key="YOUR_API_KEY")
response = client.generate_response(
{
"model": "google/gemma-3-27b:free",
"input": "Reply with: docs ready",
"temperature": 0,
}
)
print(response.get("output_text"))
```
If that works, go to the [Quickstart](./quickstart) and keep building from there.
***
## Choose your starting point
If you already have an integration or toolchain, start from the closest match instead of reading the docs from the top.
Keep the OpenAI-compatible shape, swap the base URL and key, and verify model support.
Move your existing routing setup over with parity checks and staged rollout guidance.
Start from cURL and the Responses API, then move into the endpoint reference when needed.
Use the TypeScript client as your main entry point for Responses, chat, images, audio, and more.
Start from the Python client if your integration lives in scripts, workers, backends, or notebooks.
Connect Vercel AI SDK apps to AI Stats without rebuilding your prompt or model flow.
Set up the gateway, Docs MCP, and coding-agent integrations from one place.
Jump straight to copy-paste requests and sample projects if you already know the basics.
***
## What are you trying to do?
Start with one successful text request, then learn where to read output and debug failures.
Pick the client SDK that matches your stack and start from real wrapper examples.
Set up Codex, Claude Code, OpenCode, or Docs MCP against AI Stats.
Move from OpenAI, OpenRouter, Anthropic, Vercel, LiteLLM, or another gateway.
Choose provider order, ranking, regional offers, presets, and failover behavior.
Build video, music, and batch workflows with polling, webhooks, and status retrieval.
***
## Most useful pages
The shortest path from signup to first successful request.
Base URL, auth, endpoint families, and endpoint-by-endpoint request details.
Copy-paste requests and sample projects across common workflows.
Scenario-driven recipes for routing, presets, async jobs, search, and agent workflows.
Key format, bearer headers, and common auth mistakes.
Understand failed requests quickly and fix the common causes.
***
## If you are using agents
AI Stats is set up to work well with coding agents and AI tools, not just humans reading docs in a browser.
Start here:
Let Codex, Claude Code, Cursor, or VS Code search the live AI Stats docs while they work.
Capability summary for agents that need to understand what AI Stats can do and how requests work.
Set up coding tools and editor workflows around the gateway.
Build tool-using, gateway-backed agent loops across the supported SDK languages.
***
## Explore models and providers
Use the data surfaces when you need to choose before you build:
Find models, pricing, context windows, capabilities, and provider offers in one place.
Compare performance, cost, and metadata side by side before you commit to one route.