Fixed
- Requests to Anthropic failed with a CORS error, so the connection test and every rewrite through Anthropic did not work. The extension now sends the header Anthropic requires for requests from an e-mail client.

Changed
- The popup restores the custom instruction you used last time. A "Remember the last custom instruction" checkbox under the field turns this off and clears the stored text. It is on by default.

Model lists are unchanged since 0.2.2.

Existing settings, provider keys and models are kept on upgrade. The new "remember" option starts enabled; the first instruction you type after upgrading is the first one that gets remembered.

Changed
- You can now select part or all of a mail that contains inline images and rewrite it. The images stay exactly as they are and keep their place in the text. The "allow moving images" option works for selections too.
- Selecting whole paragraphs replaces those paragraphs cleanly instead of nesting the new text inside an emptied paragraph.

Fixed
- Selecting text across a line break or a paragraph boundary no longer fails with "Selected text no longer matches the draft."

Updated model lists
- DeepSeek: deepseek-flash (default, DeepSeek V4.1 Flash), deepseek-v4-pro.
- Z.ai GLM: glm-5.3 (default), glm-5.3-flash, glm-5.3-flashx, glm-5.2.
- xAI Grok: grok-4.7 (default), grok-4.6, grok-4.5.
- Local LLMs: qwen3.6 (default), gemma4, glm-5.3-flash, deepseek-v4.1-flash, custom.
- OpenRouter: adds x-ai/grok-4.7, z-ai/glm-5.3 and deepseek/deepseek-v4.1-flash.

Existing settings are untouched: a saved provider and model stay as they are.

Selection rewrite fix and a provider model refresh.

Fixed
- Rewriting a selection that spanned a line break or a paragraph boundary failed with "Selected text no longer matches the draft." The selection is now matched regardless of whitespace, so multi-line selections rewrite as expected.
- When rewriting a selection, the rest of the draft, including any quoted thread, is sent to the model as context for tone, names and subject.

Updated model lists
- OpenAI: gpt-5.6-luna (default), gpt-6-astra, gpt-5.6-sol, gpt-5.6-terra. gpt-5.5 and gpt-5.4-mini are retired.
- Anthropic: claude-opus-5 (default), claude-fable-5-1, claude-sonnet-5, claude-haiku-4-5. claude-fable-5 is replaced by claude-fable-5-1.
- Google Gemini: gemini-3.8-flash (default), gemini-3.7-flash, gemini-3.5-flash-lite, gemini-3.1-pro-preview. gemini-3.6-flash is retired.
- OpenRouter: openai/gpt-5.6-luna (default), openai/gpt-6-astra, anthropic/claude-opus-5, anthropic/claude-fable-5.1, google/gemini-3.8-flash, x-ai/grok-4.6, z-ai/glm-5.2.

A fresh install now defaults to OpenAI with GPT-5.6 Luna. Existing settings are untouched: a saved provider and model stay as they are. A retired model that you typed in as a custom model keeps working.

rovider refresh: current models across every provider, plus two new providers.
New providers

xAI Grok (api.x.ai/v1): grok-4.6, grok-4.5, grok-4.1-fast.
Z.ai GLM (api.z.ai/api/paas/v4): glm-5.2, glm-5.1, glm-5-turbo,
glm-4.7-flash.

Both use the same chat-completions path as the existing providers, both are
covered by the shared provider behaviour tests, and both hosts are pinned in
the manifest host permissions, so no wildcard access is added.
Updated model lists

Anthropic: claude-opus-5 (default), claude-sonnet-5,
claude-fable-5, claude-haiku-4-5.
OpenAI: gpt-5.6-sol (default), gpt-5.6-terra, gpt-5.6-luna,
gpt-5.5, gpt-5.4-mini.
Google Gemini: gemini-3.7-flash (default), gemini-3.6-flash,
gemini-3.5-flash-lite, gemini-3.1-pro-preview. The Gemini 2.5 models are
removed ahead of their October 2026 shutdown.
DeepSeek: deepseek-v4-flash (default) and deepseek-v4-pro. The
retired deepseek-chat and deepseek-reasoner names are removed.
MiniMax: MiniMax-M3 (default), MiniMax-M2.7,
MiniMax-M2.7-highspeed.
OpenRouter: openai/gpt-5.6-sol (default), openai/gpt-5.6-luna,
anthropic/claude-opus-5, google/gemini-3.7-flash, x-ai/grok-4.6,
z-ai/glm-5.2, custom.
Local LLMs: qwen3.6 (default), gemma4, llama3.3, glm-4.7-flash,
custom.

If you already saved a default model that is now retired, open the options page
and pick a current one.