Open-weight models only
Loading status…

Live health for the OpenGateway /oss lane — curated open-weight models routed through Hugging Face Inference Providers. Every metric here is measured, never fabricated.

Live routes & providers

Per-subprovider throughput, TTFT, context, price & capabilities — sortable. Real HF router telemetry.

Open-weight model lanes

Curated /oss flagships — Kimi, MiniMax, Qwen, DeepSeek, GLM, GPT-OSS & more.

Subprovider ranking

Peak throughput and best TTFT per provider.
Throughput — tokens / second
Time to first token — lower is better

Components

Operational Degraded Outage
90-day uptime history lands with the rolling store. Until then we show current status only — never a fabricated bar strip.

Incidents

API surfaces

/oss
https://api.opengateway.one/oss
Primary open-weight lane. OpenAI- and Anthropic-compatible API for curated Hugging Face models plus full router passthrough.
/oss/v1
https://api.opengateway.one/oss/v1
OpenAI-compatible base URL for Codex, OpenCode and other clients that append /chat/completions.
/oss/v1/models
https://api.opengateway.one/oss/v1/models
Live model discovery — the same catalog that powers the models page.
/status/api
https://api.opengateway.one/status/api
The machine-readable status snapshot this page renders. Public, cacheable, no auth.

Hugging Face catalog sync

Proves the data above is fresh.