The cheapest AI token relay stations that we'd actually run in production. Effective price per million tokens after surcharges, payment fees and minimum top-ups.
Zero-registration free relay for open-source models. Ultra-fast (68 TPS) with 40 model variants.
One of the fastest-updating relay stations with the newest models — GPT-5.5, Claude Opus 4.7, Gemini …
Multi-provider API access beyond Claude/GPT
Trying newest relay options
Users who only need Claude and want premium
Users referred by community who want stability
Claude power users who want single-provider focus
Speed-sensitive Claude users
Trying before committing — ¥4 free is good for testing
Codex budget users
Codex users willing to pay for reliability
Casual, low-budget users
Speed-sensitive users willing to test new options
Budget-friendly Claude API relay with competitive pricing. Popular among developers for coding …
Users who want transparency via status page
Users wanting to test Claude Code options
Users who need access to many different models
Users who want one API key to rule them all
Users who prioritize channel quality over price
Claude Code users who need reliable MAX quality
Enterprise users who need zero dilution
Budget-conscious developers, Codex users
Developers who value documentation and stability
Multi-model users who want low commitment
Pure budget testing, not production
Budget users willing to accept quirks
Domestic direct-connect relay — no VPN needed for Chinese users. Simple pay-per-use pricing with …
Budget testing alongside established stations
Claude Code users wanting variety
Users wanting subscription model vs pay-as-you-go
New users wanting to test with minimal commitment
Bleeding-edge testers only
Testing additional relay options
Chinese-speaking developers
Free forever API relay with strict no-logging policy. Handles 1B+ tokens daily with open-source …
The 8 cheapest AI token relay stations we’d actually run in production — ranked by effective $/1M input tokens after surcharges, payment fees and minimum top-ups. We exclude any platform with a security rating below A or a measured outage over 30 minutes in the last 30 days, no matter how cheap.
Effective price per million tokens after surcharges (payment fees, minimum top-ups, markup), divided by the safety rating. We exclude any relay below an A security score.
No. The cheapest relays often skip security audits or hide markup in surcharges. The platforms here are filtered to a minimum A security score and >95% measured uptime.