LLM inference APIs you can pay per request with x402

Several providers resell model inference over x402, so an agent can call a model without an account or API key. Prices vary a lot by model and output length.

Data from the crawl of 2026-11-19 02:23 UTC. An endpoint matches when its name, URL, category or description contains llm. It is LIVE when it answered with an HTTP 402 on its latest probe.

The market right now

LIVE matching endpoints580 from 264 services
With a valid 402 body575
Price per call (USDC)min 0.0001 · median 0.0088 · 90th percentile 0.08
Score 90+ (of 100)519 · median score 100
Median probe latency323 ms
Top chainsBase 554, Solana 199, Polygon 118, Arbitrum 88, nano:mainnet 32, Optimism 18

Top 10 by the router's ranking

EndpointUSDCChainsScoremsHistory
Agent Extraction API /api/free-tiers
freetier-agent-api.pages.dev
0.001Base10027probes
x402-data-api.onrender.com /api/calc
x402-data-api.onrender.com
0.001Base10037probes
x402-data-api.onrender.com /api/calc/dates
x402-data-api.onrender.com
0.001Base10038probes
WebberSites x402 Data API /api/calc
api.webbersites.com
0.001Base10059probes
tlacuilo-jet.vercel.app /api/v1/web/read
tlacuilo-jet.vercel.app
0.001Solana, Base10065probes
forge-tools.dentreprise.workers.dev /api/json-schema-validate
forge-tools.dentreprise.workers.dev
0.001Base10075probes
JSON Schema Validator /api/validate-json
keyronne.com
0.001Base10079probes
Quartermaster Random Numbers /v1/random
quartermaster.surewhynot.app
0.001Base, Solana10089probes
Degenscan Intel /v1/ai/cheapest
intel.degenscan.io
0.001Base, Solana100117probes
x402-random-bytes /random
x402-catalog-history.ai-business-surface.workers.dev
0.001Base100128probes

Ranking: valid 402 body first, then score, price, latency. The free /route returns the first 3. /route/paid returns up to 25 with payTo, asset and networks.

How to choose

Do it from code

Free best match + top 3:

curl -s -X POST https://router-production-ab89.up.railway.app/route -H 'content-type: application/json' -d '{"q": "llm"}'

Full ranked list with a ready-to-pay pick (0.01 USDC, x402):

# paid_client.py from https://router-production-ab89.up.railway.app/examples/paid_client.py
import asyncio
from paid_client import route_paid
print(asyncio.run(route_paid('llm')))

More: LangChain, CrewAI, MCP, AgentKit guides · browser demo

Prices: /route/paid 0.01 USDC, /check/paid 0.002 USDC, /route free. x402 v2 on Base (eip155:8453) or Solana. Machine-readable: OpenAPI · /.well-known/x402 · llms.txt