Sign in to Devin Desktop with ChatGPT and set Fusion to GPT-6 Astra + SWE-2
Upgrade Devin Desktop to v3.10.48, sign in with an eligible ChatGPT Plus or Pro plan so GPT requests bill to that plan, and use Fusion with GPT-6 Astra as lead and SWE-2 as sidekick. Use the new LS memory/CPU profile commands when filing Desktop issues.Why this loop
v3.10.48 lets you sign in to Devin with ChatGPT and bill eligible GPT usage in Devin Desktop to a ChatGPT Plus or Pro plan. The release’s value path is explicit: choose a GPT model so those requests hit the ChatGPT plan, and for best value run Fusion with GPT-6 Astra as the lead and SWE-2 as the sidekick, following the in-product “Use your ChatGPT plan” setup for eligible models. The same build adds Command Palette diagnostics—Devin: Save Language Server Memory Profile and Devin: Save Language Server CPU Profile—that write a .zip for support. It also stops long/noisy shell output from growing Agent-window memory, keeps every folder of a converted multi-folder workspace when a new session starts without a folder pick, shrinks the local settings file for accounts with many models, and unstalls typing/copy-paste on the extension host. Upgrade first so auth and model routing are not fighting already-fixed Desktop bugs.
Proposed actions
- Upgrade Devin Desktop to v3.10.48 from https://windsurf.com/changelog#v3-10-48 before changing auth or model routing.
- In Devin Desktop, sign in with ChatGPT using an eligible ChatGPT Plus or Pro plan so GPT model requests are billed to that plan; complete the in-product “Use your ChatGPT plan” setup and confirm eligible models.
- For GPT-billed Devin Desktop sessions, set Fusion with GPT-6 Astra as the lead and SWE-2 as the sidekick.
- If a language server is slow or leaking, open the Command Palette and run “Devin: Save Language Server Memory Profile” and “Devin: Save Language Server CPU Profile”, then attach the saved diagnostic .zip to support.
- On a converted multi-folder workspace, start the next Agent session without picking a folder so Devin Desktop keeps every folder instead of only the first one.
Agent prompt
Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI costCascade agent task
DevAgentRadar → Windsurf Cascade
Goal: analyze and pilot this coding release update in my project.
Context
Assistant: Windsurf Proposal: Sign in to Devin Desktop with ChatGPT and set Fusion to GPT-6 Astra + SWE-2 Summary: Upgrade Devin Desktop to v3.10.48, sign in with an eligible ChatGPT Plus or Pro plan so GPT requests bill to that plan, and use Fusion with GPT-6 Astra as lead and SWE-2 as sidekick. Use the new LS memory/CPU profile commands when filing Desktop issues. Primary source: https://windsurf.com/changelog#v3-10-48
Why it matters
v3.10.48 lets you sign in to Devin with ChatGPT and bill eligible GPT usage in Devin Desktop to a ChatGPT Plus or Pro plan. The release’s value path is explicit: choose a GPT model so those requests hit the ChatGPT plan, and for best value run Fusion with GPT-6 Astra as the lead and SWE-2 as the sidekick, following the in-product “Use your ChatGPT plan” setup for eligible models. The same build adds Command Palette diagnostics—Devin: Save Language Server Memory Profile and Devin: Save Language Server CPU Profile—that write a .zip for support. It also stops long/noisy shell output from growing Agent-window memory, keeps every folder of a converted multi-folder workspace when a new session starts without a folder pick, shrinks the local settings file for accounts with many models, and unstalls typing/copy-paste on the extension host. Upgrade first so auth and model routing are not fighting already-fixed Desktop bugs.
Suggested actions
-
Upgrade Devin Desktop to v3.10.48 from https://windsurf.com/changelog#v3-10-48 before changing auth or model routing.
-
In Devin Desktop, sign in with ChatGPT using an eligible ChatGPT Plus or Pro plan so GPT model requests are billed to that plan; complete the in-product “Use your ChatGPT plan” setup and confirm eligible models.
-
For GPT-billed Devin Desktop sessions, set Fusion with GPT-6 Astra as the lead and SWE-2 as the sidekick.
-
If a language server is slow or leaking, open the Command Palette and run “Devin: Save Language Server Memory Profile” and “Devin: Save Language Server CPU Profile”, then attach the saved diagnostic .zip to support.
-
On a converted multi-folder workspace, start the next Agent session without picking a folder so Devin Desktop keeps every folder instead of only the first one.
Config surfaces this release may change
- MCP servers — check your repo before applying
After you finish
Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.
Tell the human: open https://devagentradar.com/proposals/windsurf-v3-10-48-sign-in-to-devin-desktop-with-chatgpt-and-set-fusion-t and mark Applied, Skipped, or Failed. Proposal id: 73aee6a8-fd11-4332-a679-9210aa24d7be
Instructions:
- Provide a brief breakdown of potential workflow improvements
- Propose localized updates without touching unrelated logic
- Verify changes with relevant tests before concluding
Confirm receipt of task and wait for my instruction to begin.
Your loop
This browser · no sign-in · not shared as “you”After you run the prompt
Only you can mark this. Agents cannot write your loop.
Your decision stays on this device. A public tally appears after a few votes.