▸ status usage
new: genuinely free AI models are live — try surp/free + see live budgets · token-gating prototype · vote on SRP

status

systemctl status surp-gateway

3578
total requests served
90
requests (24h)
$3.02
x402 USDC settled
$5.30
API-key usage tracked
16
unique wallets

system health

resolverok
facilitatorhttps://facilitator.payai.network
cached markets415
last request5h ago
networkeip155:8453
markup500 bps (5%)

cache engine

22.22%
historical exact-cache hit rate · 27 eligible lookups
167
tokens not recomputed
0
live cached answers · 15 min TTL
0
sticky route reuses
exact response TTL900s
cache-hit price$0.001 per request
sticky route window300s
price tolerance30% above live cheapest

Only deterministic, non-streaming, tool-free requests use exact-response caching. Prompts are SHA-256 fingerprinted and are never stored. how the cache engine works »

cache flywheel rewards (SRP)

$1.1335
rebate pool backing
1386
SRP outstanding
9
reward holders
0.081779¢
value per SRP

Cache writers and readers earn off-chain SRP points backed by an earmarked share of gateway revenue. This is a simulation ledger for a future RevNet/Juicebox deployment — no token exists on-chain yet.

sponsored free tier

89
free requests remaining today
16,599
free tokens remaining today
11
free requests served today
4612ms
average free latency

surp/free is treasury-sponsored with no wallet or payment required. live eligible models, budgets, and methodology »

top combos

comborequests
direct/deepseek-v4-flash-0731496
direct/deepseek-v4-flash490
direct/glm-5.2480
direct/deepseek-v4-pro480
direct/gpt-5.6-luna479

top models routed

modeltimes routed
deepseek-v4-flash805
deepseek-v4-flash-0731496
glm-5.2480
deepseek-v4-pro480
gpt-5.6-luna479

live data from GET /api/stats. stats are best-effort — a logging failure never blocks a request.