A drop-in OpenAI-compatible chat proxy that repairs broken upstream streams and enforces per-client budgets and rate limits. Point any OpenAI client at it — your models stay upstream, your spend stays predictable.
Drop-in endpoint for chat completions — point your existing SDK or IDE client at Streamproof and it just works.
Upstream truncations, dropped chunks, and malformed SSE are detected and repaired so clients see a clean completion.
Monthly spend caps per client key. When a budget is exhausted, requests are cleanly rejected instead of silently burning money.
Per-key requests-per-minute throttling with predictable backoff — no noisy-neighbor problems on shared upstreams.
Every request is metered per key: tokens, cost, latency, and errors — visible in the Streamproof dashboard.
Same request in, same cost out. Billing is computed server-side before the upstream call completes.
Swap your base URL to https://streamproof.aetherin.org/api and use your Streamproof client key as the bearer token:
curl https://streamproof.aetherin.org/api/chat \
-H "Authorization: Bearer $STREAMPROOF_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-4o-mini",
"stream": true,
"messages": [
{"role": "user", "content": "Say hello"}
]
}'Any OpenAI SDK works too — set baseURL to https://streamproof.aetherin.org/api and apiKey to your client key.
Streamproof client keys are provisioned by an administrator — each key carries its own monthly budget and RPM limit. Keys are issued on request — contact support@aetherin.app to have a key issued for your use case. Once you have a key, the Streamproof dashboard shows your usage, budget remaining, and rate-limit headroom.