# Synap — The AI of the People (synap.linkrra.com)

> Synap is Linkrra's OpenAI-compatible inference API for open models, self-hosted on
> GPUs Linkrra rents and operates or bought through zero-data-retention providers,
> with the lowest published price anywhere on qwen3-coder-30b (checked 2026-10-08). Agents can discover models, estimate costs and
> pay per call autonomously.

Contact: weare@linkrra.com

## For AI agents: how to buy inference here
- Models + live prices (no key): GET https://pool.linkrra.com/v1/models
- Chat (OpenAI-compatible): POST https://pool.linkrra.com/v1/chat/completions
  with `Authorization: Bearer <key>` — self-serve keys at https://synap402.com
- MCP server: https://synap402.com/mcp (also pool.linkrra.com/mcp) —
  3 free tools (list models / get model / estimate cost) + billed chat.
  Source: https://github.com/jimmyplup/synap-mcp · Registry: com.linkrra/synap
- No-account pay-per-call via x402 (USDC on Base): https://synap402.com/.well-known/x402
- Top up credits by card, TON, x402 or PRL: https://synap402.com/account

## Self-hosted models (GPUs we rent and operate, zero retention)
qwen3-coder-30b · gpt-oss-20b · gemma-4-26b-a4b-it · qwen/qwen3-30b-a3b — authoritative prices at
GET https://pool.linkrra.com/v1/models?format=openrouter. Brokered frontier
models (GLM, DeepSeek, Kimi and more) are available behind the same API.

## Pricing model
Pay-as-you-go credits (a credit is a unit of money, not a token): each request
costs the per-token USD price /v1/models advertises for the model that served
it. Plans from 9/mo at https://synap402.com/pricing.