Twitxy Saved

Lingming Zhang

@LingmingZhang

Associate Professor @plfmse @IllinoisCS.
Enjoy breaking, fixing, and synthesizing software.
SE | PL | FM | LLM4Code
Champaign, IL lingming.cs.illinois.edu Joined May 2012
459 Following 3.6K Followers 525 Posts

Latest posts 5

Lingming Zhang @LingmingZhang
90.5% on SEC-bench Pro๐Ÿคฏ๐Ÿคฏ๐Ÿคฏ
Sundar Pichai Verified account @sundarpichai 2d
Gemini 4 Neon: frontier-level intelligence with fast end-to-end task completion across complex software engineering, knowledge work, and science workflows.

Begins rolling out today in the Gemini API in @GoogleAIStudio, Gemini for business, for paid Google AI Pro and Ultra subscribers in Google @Antigravity, and for Google AI Pro and Ultra subscribers in the @GeminiApp.
View quoted post on X
0 replies 0 reposts 25 likes 941 views View on X
Lingming Zhang @LingmingZhang
We look forward to your submissions on #LLM4Code!!
Saikat Dutta @saikatdutta2012 Oct 14
๐Ÿ“ฃ Announcing the 4th workshop on #LLM4Code, co-located with @ICSEConf 2027 in Dublin, Ireland ๐Ÿ‡ฎ๐Ÿ‡ช!

๐ŸŽฏ We are calling for submissions:

๐Ÿ”— Website: llm4code.github.io
๐ŸŒŸ Deadlines:
โ€ข Submission: Nov 13, 2026 (23:59 AoE)
โ€ข Notif.: Dec 11, 2026
โ€ข Camera-ready: Jan 29, 2027
View quoted post on X
Lingming Zhang @LingmingZhang
Tired of comparing LLMs across proprietary, apples-to-oranges agent scaffolds for SWE tasks?

๐Ÿ“ข๐Ÿ“ข Introducing a unified leaderboard: all models are evaluated using live-SWE-agent, the first live software agent that self-evolves on the fly.

๐Ÿ”ฅ Opus 4.5 + live-SWE-agent hits 79.2% on SWE-bench Verified (Top-1), with Gemini close behind! Moreover, live-SWE-agent outperforms all open software agents and even approaches the best proprietary scaffold used by the Claude team!

๐Ÿš€ We hope to offer an open, unified, and powerful SWE scaffold for fair, apples-to-apples comparison of future model releases. Check it out: github.com/OpenAutoCoder/โ€ฆ
5 replies 11 reposts 63 likes 5.4K views View on X
Lingming Zhang @LingmingZhang
๐Ÿคฏ๐Ÿคฏ๐Ÿคฏ Gemini 3 Pro + Live-SWE-agent hits 77.4% on SWE-bench Verified, beating ALL existing models, including Claude 4.5!!

๐Ÿค– Live-SWE-agent is the first live software agent that autonomously self-evolves on the fly โ€” and it even outperforms the manually engineered scaffold used by the Gemini 3 Pro team (76.2%)
33 replies 66 reposts 470 likes 115K views View on X
Lingming Zhang @LingmingZhang
๐Ÿš€ Introducing Live-SWE-agent:

๐Ÿค– An autonomous coding agent that self-evolves on the fly while solving real-world issues.
โœจ Simplistic design: no offline training, no heavy workflows.

๐Ÿ‘‡ Surprising gains:

๐Ÿ“Œ 75.4% on SWE-bench Verified
๐Ÿ“Œ 45.8% on SWE-Bench Pro (new SOTA!)