4.3.12 2026-10-15

Extension release: a ChatGPT Plan provider, contributed by @CoderLambert (#540). twinny-server refuses the new provider kind, which needs the editor, and masks more secrets in its logs.

• ChatGPT Plan provider. Completions and chat on the models your ChatGPT account can use, without an API key. Pick ChatGPT Plan in the provider presets, Sign in opens ChatGPT's sign-in in the browser, and the model list fills from your account. The sign-in is kept in VS Code's secret storage and survives restarts; Sign out ends it and stops any request in flight. A Reasoning effort setting (model default, none to maximum) trades answer quality for speed; lower is better for completions. Requests go to OpenAI's public Responses API with store: false. Embeddings are not offered. Expect hosted latency for completions (about 2–4 seconds), not a local model's.
• Fixed: a credential the secret shield put back into a reply could be written to the log. Log lines now go through the shield's secret detection too, in the extension's output channel and twinny-server's stderr.