Local Models Get Time to Finish, in Either Provider Shape

Fixed

• A local model now gets time to finish on the Anthropic provider too. Point the Anthropic provider's base URL at a model on this computer or your local network — 1LocalAITool answers in that shape as well — and it now waits up to ten minutes, the same as a custom OpenAI-compatible endpoint since 1.5.0, instead of failing after 45 seconds. Hosted Anthropic keeps its usual limit.

• Long pieces from a model on your own machine no longer fail halfway. If you are coming from 1.4.0: a custom OpenAI-compatible endpoint on this computer or your local network — 1LocalAITool, LM Studio, Ollama, or one reached over Tailscale — gets up to ten minutes to answer instead of 45 seconds, so a campaign or article from a model running on a CPU finishes instead of timing out.