You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, I'm new with this tool and tried API connection with Gemini as well as with Groq.
With Gemini after given the API Code and click on test. Failur 404
With Groq: API Error: The model llama-3.3-70b-versatile does not exist or you do not have access to it.
I do not see how I could change the model anywhere.
👋 Welcome and thanks for opening your first issue!
We appreciate you taking the time to contribute or report a problem. Someone will review this as soon as possible. In the meantime, please make sure you've included all relevant details (like logs, steps to reproduce, or environment info).
Thanks for the detailed report! 🙏 Both problems are addressed in the new v1.2.4 release:
Gemini (404 on Test): The Gemini API works fine — it just needs a proper API key. A 404 usually means the key wasn't copied completely, or the Generative Language API isn't enabled for the Google Cloud project the key belongs to. Please double-check the setup guide: Documentation → Setup Guide
Groq (llama-3.3-70b-versatile does not exist): The model we were using was shut down in August, which is exactly why you got that error. It has now been updated to the new model we tested: openai/gpt-oss-120b.
Thanks @benjamindahl — you're right, and I can now reproduce the exact cause. 🙏
Why the 404 happens
Your curl to .../v1beta/models?key=... succeeds because listing models isn't restricted. But generateContentis. On September 18, 2026 Google changed the rules:
"we are limiting access to the 2.5 models to users who have actively used them in the past. For any new projects, use our latest models: 3.5 Flash-Lite or 3.8 Flash."
So a brand-new project + API key (exactly what you created) gets:
404 NOT_FOUND
models/gemini-2.5-flash is no longer available to new users.
even though models?key= still lists it. It's not a key problem or a plan problem — the restriction is tied to the project's past usage history, which is why it works for some users (their old project is whitelisted, even with a freshly generated key) and not for you. gemini-2.5-flash also shuts down for everyone on Oct 16, 2026.
Fixed in v1.2.5
I've just released v1.2.5, which:
Defaults Gemini to gemini-3.5-flash-lite (available to new projects)
Adds a Gemini Model selector, plus a Load Available Models button that queries your key and lists exactly which models it can use
Adds an optional Model override for the other providers too
Handles the Gemini 3.x thinking config correctly (thinkingLevel instead of the now-invalid thinkingBudget)
Shows the API's real error message (e.g. "no longer available to new users") instead of a bare 404
Install: Thunderbird → Add-ons and Themes → gear icon → Install Add-on From File... → pick the .xpi.
Then open Settings → Gemini Model and either keep gemini-3.5-flash-lite or click Load Available Models to pick from the list. Please let me know if it works for you now!
Activity
github-actions commentedon Oct 9, 2026
👋 Welcome and thanks for opening your first issue!
We appreciate you taking the time to contribute or report a problem. Someone will review this as soon as possible. In the meantime, please make sure you've included all relevant details (like logs, steps to reproduce, or environment info).
Nigel1992 commentedon Oct 10, 2026
Thanks for the detailed report! 🙏 Both problems are addressed in the new v1.2.4 release:
Gemini (404 on Test): The Gemini API works fine — it just needs a proper API key. A 404 usually means the key wasn't copied completely, or the Generative Language API isn't enabled for the Google Cloud project the key belongs to. Please double-check the setup guide: Documentation → Setup Guide
Groq (
llama-3.3-70b-versatiledoes not exist): The model we were using was shut down in August, which is exactly why you got that error. It has now been updated to the new model we tested:openai/gpt-oss-120b.Download the fixed
.xpi(ready to install): https://github.com/Nigel1992/AutoSort-Plus/releases/tag/v1.2.4Install: Thunderbird → Add-ons and Themes → gear icon → Install Add-on From File... → pick the
.xpi.Feel free to reopen or leave a comment if you still hit any issues — happy to help!
benjamindahl commentedon Oct 14, 2026
Same problem here
404 Error on Test with Gemini. I'm having a Pro plan and created a new project and a new API Key in Google AI Studio
Checked it with curl "https://generativelanguage.googleapis.com/v1beta/models?key=[MY-API-KEY]" which works
benjamindahl commentedon Oct 14, 2026
I can see this error in the AI Studio Console
benjamindahl commentedon Oct 14, 2026
It should also find Gemini 2.5 flash
Want to add an idea here… Let the user choose the model
Nigel1992 commentedon Oct 14, 2026
Thanks @benjamindahl — you're right, and I can now reproduce the exact cause. 🙏
Why the 404 happens
Your
curlto.../v1beta/models?key=...succeeds because listing models isn't restricted. ButgenerateContentis. On September 18, 2026 Google changed the rules:So a brand-new project + API key (exactly what you created) gets:
even though
models?key=still lists it. It's not a key problem or a plan problem — the restriction is tied to the project's past usage history, which is why it works for some users (their old project is whitelisted, even with a freshly generated key) and not for you.gemini-2.5-flashalso shuts down for everyone on Oct 16, 2026.Fixed in v1.2.5
I've just released v1.2.5, which:
gemini-3.5-flash-lite(available to new projects)thinkingLevelinstead of the now-invalidthinkingBudget)404Download (ready to install): https://github.com/Nigel1992/AutoSort-Plus/releases/tag/v1.2.5
Install: Thunderbird → Add-ons and Themes → gear icon → Install Add-on From File... → pick the
.xpi.Then open Settings → Gemini Model and either keep
gemini-3.5-flash-liteor click Load Available Models to pick from the list. Please let me know if it works for you now!benjamindahl commentedon Oct 15, 2026
Works now!
Nigel1992 commentedon Oct 15, 2026
@Jannus88 Can you please confirm your issue has been fixed aswell in the latest release?
If so, please close the issue.