$relay-compare --scan --all --daily
Every AI API relay,
benchmarked 24/7.
We probe 176+ relay stations every day for quality, latency, uptime and trust — so you can route Claude, GPT, Gemini and DeepSeek traffic through gateways that actually hold up.
probe.log
// last scan 2026-11-01 16:42 SGT ▸ probing 176 relays across 8 providers ▸ claude-sonnet-4-6 ✓ reachable 172 ▸ gpt-5 ✓ reachable 167 ▸ gemini-2.5-pro ~ partial 150 ▸ trust classified ✓ 176/176 done in ~4m 12s
176+
Relays tracked
8
Model providers
24h
Scan cadence
100%
Reachable
176 / 176 relays
| Fidelity | Trust | Models | ||||||
|---|---|---|---|---|---|---|---|---|
| 01 | BUZZ GLOBALhttps | 7.1 | unverified | 72 ms | 96.7% | medium | Anthropic (Claude)OpenAI (GPT)+3 | open |
| 02 | Avian GLOBALhttps | 7.1 | real | 89 ms | 96.7% | medium | Anthropic (Claude)OpenAI (GPT)+2 | open |
| 03 | Featherless GLOBALhttps | 7.1 | unverified | 91 ms | 96.7% | medium | OpenAI (GPT)DeepSeek+3 | open |
| 04 | OpenRouter GLOBALhttps | 7.1 | unverified | 99 ms | 96.7% | medium | Anthropic (Claude)OpenAI (GPT)+5 | open |
| 05 | ClaudeCN CNhttps | 7.1 | unverified | 105 ms | 96.7% | medium | Anthropic (Claude)OpenAI (GPT)+1 | open |
| 06 | zen-ai.top CNhttps | 7.1 | unverified | 64 ms | 96.7% | medium | OpenAI (GPT) | open |
| 07 | API Management CNhttps | 7.1 | unverified | 77 ms | 96.7% | medium | OpenAI (GPT) | open |
| 08 | apirouter.ai GLOBALhttps | 7.1 | unverified | 72 ms | 96.7% | medium | open | |
| 09 | CloseAI CNhttps | 7.1 | unverified | 59 ms | 96.7% | medium | Anthropic (Claude)OpenAI (GPT)+1 | open |
| 10 | PoloAPI-直连官转中转API CNhttps | 7.1 | unverified | 59 ms | 96.7% | medium | Anthropic (Claude)OpenAI (GPT)+6 | open |
| 11 | Requesty GLOBALhttps | 7.1 | unverified | 84 ms | 96.7% | medium | Anthropic (Claude)OpenAI (GPT)+2 | open |
| 12 | Machine Learning Models and Infrastructure GLOBALhttps | 7.1 | unverified | 90 ms | 96.7% | risky | Anthropic (Claude)Google Gemini+4 | open |
| 13 | SambaNova GLOBALhttps | 7.1 | unverified | 52 ms | 96.7% | risky | OpenAI (GPT) | open |
| 14 | Production-scale AI with Ray GLOBALhttps | 7.1 | unverified | 71 ms | 96.7% | risky | Meta Llama | open |
| 15 | siliconflow-com GLOBALhttps | 7.1 | unverified | 106 ms | 96.7% | risky | OpenAI (GPT) | open |
| 16 | nextweb-dev GLOBALhttps | 7.1 | unverified | 29 ms | 96.7% | risky | open | |
| 17 | 辣椒炒肉 GLOBALhttps | 7.1 | unverified | 60 ms | 96.7% | risky | open | |
| 18 | token-ai-com GLOBALhttps | 7.1 | unverified | 63 ms | 96.7% | risky | open | |
| 19 | Modal: The platform for production AI GLOBALhttps | 7.1 | unverified | 68 ms | 96.7% | risky | open | |
| 20 | chatgptid-com GLOBALhttps | 7.1 | unverified | 76 ms | 96.7% | risky | open | |
| 21 | Zensors GLOBALhttps | 7.1 | unverified | 95 ms | 96.7% | risky | open | |
| 22 | 7.1 | unverified | 98 ms | 96.7% | risky | open | ||
| 23 | PackyAPI (PackyCode) CNhttps | 7.0 | unverified | 151 ms | 96.7% | medium | Anthropic (Claude)OpenAI (GPT)+1 | open |
| 24 | ggwk1.online CNhttps | 7.0 | unverified | 172 ms | 96.7% | medium | Anthropic (Claude)OpenAI (GPT)+1 | open |
| 25 | RightCode CNhttps | 7.0 | unverified | 169 ms | 96.7% | medium | Anthropic (Claude)Google Gemini+1 | open |
showing 1–25 of 176
page 1 / 8
Fidelity column is a placeholder ranking — real probe-based authenticity detection is on the roadmap. See Methodology.