Lovable ships with GPT-6 Sol, 6-12% better on 0-to-1 benchmark
OpenAIDevs · x · 2026-09-23
Lovable announces it now builds with GPT-6 Sol. On its 0-to-1 building benchmark the model scores 6-12% higher than GPT-5.6 Sol at every effort level, with the biggest gains in following the brief (10-15% higher intent alignment) and design fundamentals (8-10% higher).
More from coding & agent
- ReFigBench paper: same model scores swing on identical tasks across Claude Code and Codex harnesses — omarsar0 · 2026-09-23
- NVIDIA's Skill2Env turns 3.4k Agent Skills into 8k RL environments, boosting Qwen-27B by 4.7 points — burny_tech · 2026-09-23
- 93% of new Netlify projects are agent-created, so it built Git storage for millions — jasonkneen · 2026-09-23
- Opus rebuilds San Francisco in Unreal Engine, with Jev powering everything in it — MatthewBerman · 2026-09-23
- Pulumi Engineer's 10 Coding Agent Tips: Move Instructions Out of Chat Into Files and Hooks — rseroter · 2026-09-23
- The Virtual Biotech: 37,000 AI agents organized for drug discovery — virusxp · 2026-09-23