Overall status · 🧊
Cooling down
We intentionally slowed public traffic during the refresh.
Robin AI · Beta 🚧
Thank you 🙌 for pushing our beta to its limits.
We appreciate every test, bug report, and idea you shared while we were live in beta. The public API plane is intentionally offline while we polish the next release of Robin AI—a developer-first platform for high-performance AI models.
🔐 When we return, the API will live on its own dedicated endpoint domain for added isolation, stability, and faster routing across regions.
Healthy systems, just taking a breath 🫧
The last health snapshot looked solid, but we intentionally powered down the public surface to fold in your beta notes. Endpoints may intermittently reject traffic while we rewire things, so please treat the pause as a maintenance window rather than a green light to keep shipping calls.
Infrastructure · 🛡️
Healthy
GPU pool tuned, routing layer getting extra guardrails.
Observability · 👀
Paused stream
Live dashboards resume once the console reopens.
Model lineup returning soon 💪
These are the OpenAI-compatible workhorses you relied on during beta. They're offline with the rest of the stack for now, but they'll headline the refreshed console the moment we reopen traffic.
General purpose · 🌌
Kimi‑K2‑Thinking
1 trillion parameters tuned for long-form reasoning, planning, and strategy-heavy prompts.
Top choice for async agents and complex workflows.
Compact reasoning · ⚡
DeepSeek‑v3.1
671 billion parameters delivering high throughput with surprisingly grounded answers and efficient context usage.
Great middle ground for production assistants.
Code generation · 💻
Qwen3‑Coder
480 billion parameters purpose-built for repo-scale understanding and precise code edits.
Beta teams leaned on it for PR reviews and scaffolds.
Tokens burned during beta 🔥
Across the final six beta days, 76+ API keys generated 1,520,866 tokens. After applying our internal traffic weighting, that becomes 111,023,218 virtual tokens worth of experimentation—plenty of signal to guide this refresh.
Days counted
20–25 Nov
Last active stretch before maintenance.
Raw tokens
Summed from the request ledger.
API keys engaged
Final burst of testers before the pause.
Weighted total
Internal weighted load proxy.
Weighting helps us simulate full production load without keeping the public endpoints live.
What's coming back even better ✨
Infra refresh
Faster inference across every hosted model thanks to upgraded GPUs and smarter routing.
New analytics
Real-time usage insights so teams can understand spend, latency, and success rates at a glance.
Expanded API surface
More OpenAI-compatible endpoints plus streaming and fine-tuning workflows.
API base URL · 🌐
https://robin.sdad.pro/api/v1
Temporarily offline during maintenance.
Next status update · 🗓️
Early December
Questions? · ✉️
contact@sdad.pro
Built for developers, by developers 🛠️
Robin AI is a community-driven effort to keep powerful models accessible. We are rebuilding with your beta feedback in mind so you can ship faster without watching the meter.
Keep checking this page for the next status drop.