Agent developer seeks failover patterns after single-provider outage killed every run
Arkupathak · reddit · 2026-09-27
A developer describes how a single-provider setup caused every agent run to fail when that provider degraded, and is now designing automatic fallback: if model A errors or times out, retry on an equivalent model elsewhere. He asks three open questions: fall back to the same model on another host or a different model (prompts don't always carry over); build it yourself or use a gateway (LiteLLM, Portkey, llmapi.ai all advertise fallbacks); and how to keep fallbacks from quietly running up costs by landing on pricier models.
More from coding & agent
- Baseten's Guide: Evaluate Open Models for Coding Agents on Your Own PRs — baseten · 2026-11-16
- Google Research open-sources PaperOrchestra, a multi-agent framework for writing AI papers — google-research · 2026-11-16
- ffmpeg-skill gives coding agents local video editing via 42 FFmpeg tools and MCP — kajisho5 · 2026-11-16
- Agents Now Reply to Slack Messages on Their Own, Resolving a Prod Incident Ghost-to-Ghost — iamrobotbear · 2026-11-16
- Engineer's Interview Prep Method: Use Claude to Find Which Answers Undersell Your AI Work — brandon_galang · 2026-11-16
- DeepSeek open-sources its app frontend "chat-demo", already at 3.9k stars — NielsRogge · 2026-11-16