Twitxy Saved

Aashna Garg

@aashna

Github Copilot | Former Stanford
Seattle Joined February 2009
257 Following 285 Followers 365 Posts

Latest posts 5

Pinned post

Aashna Garg @aashna
Super proud to share Copilot Fusion on top of TerminalBench4.1 at fraction of cost, announced by @satyanadella at Microsoft Ignite today! Surreal to attend this live. Tune in to our talk on Copilot Fusion later tomorrow at #MSIgnite !
Mario Rodriguez Verified account @mariorod1 3d
You shouldn't have to pick a model for every task. Copilot Fusion does it for you.

As @SatyaNadella just shared at #MSIgnite, Copilot Fusion now ranks as the #1 coding agent on Terminal-Bench, scoring 5 points ahead of Claude Code at less than half the cost.
View quoted post on X
Aashna Garg @aashna
My #MSIgnite talk is live: Designing an agent harness that orchestrates models. Copilot Fusion routes models as a portfolio — and it is #1 on Terminal-Bench 4.1: 71.2%, +5.4 pts ahead, at a fraction of the cost. Watch: youtu.be/1kpeIbdAfoQ
@github @Microsoft
www.youtube.com Designing an agent harness that orchestrates models | BRK280 Choosing a model is becoming less important than deciding how and w...
Aashna Garg @aashna
Today at #GitHubUniverse we launched Copilot Fusion 🚀

Proud to have led the research behind it. Fusion orchestrates Anthropic + OpenAI models per task:

🏆 #1 on Terminal-Bench 4.0
💰 ~half the cost of the top agent
⚡ ~half the tokens

Grateful to this incredible team 🙏
GitHub Verified account @github Oct 28
Frontier performance at a fraction of the cost.

In the @ArtificialAnlys Coding Agent Index, Copilot Fusion Max scored within two points of Claude Code with Opus 5.5 Max at about half the API cost per task.

Get 30% off through Nov. 30. Try it now. ⬇️ docs.github.com/copilot/concep…
View quoted post on X
Aashna Garg @aashna
We are now available in Github App and VSCode!! Keep the feedback coming!
GitHub Verified account @github Sep 30
👋 Project HydraFusion is now available in the GitHub Copilot app and @code.

Choose it like any other model. Behind the scenes it routes your task across multiple models to draft, critique, revise, or escalate, then hands you one result. github.blog/changelog/2026…
View quoted post on X