Grok 4.5 Takes Second Place on SWE Leaderboard
elonmusk · x · 2026-07-11
Grok 4.5 scored Pass@1 51.2% ±6.0 on the real-world software engineering benchmark APEX-SWE, ranking second behind Fable 5's 65.5% ±6.2.
In specific subcategories, it ranked first in Integration tasks with 65.0% Pass@1 and second in Observability with 37.3% Pass@1. The post also notes that these multi-step build and diagnostic/debugging tasks perfectly align with the agentic coding workflows Grok 4.5 targets. Compared to Grok 4's 21.0%, this represents a 30.2 percentage point improvement within a year.
Related event: Grok 4.5 Ranks No. 2 on APEX-SWE(2 posts)→
More from coding & agent
- Hugging Face wraps 6-month Training Agents series with open recordings and code — SergioPaniego · 2026-10-26
- Agent verified 'success' after double-voting: a lesson in harness vs model bugs — TejasKumar_ · 2026-10-26
- diginomica: agentic automation hype is overdone, and 'The Hackening' of enterprise security has begun — jonerp · 2026-10-26
- 'Vibe verification': ~60% of AI-written formal specs prove the wrong property — basedjensen · 2026-10-26
- Devin launches Security suite powered by Claude Mythos with $100M+ credits — DevinAI · 2026-10-26
- HackBench injects reward hacks into agent trajectories to stress-test monitors — DanielKhashabi · 2026-10-26