Simon Willison reviews key 2026 LLM developments, highlighting November 2025 releases Claude Opus 4.5 and GPT-5.1 as incremental but meaningful upgrades that made coding agents reliably usable day-to-day. He also tests models on an SVG pelican-on-bicycle benchmark and notes early activity in an obscure GitHub repo called Warelay.
Background
The large language model landscape continues to evolve rapidly, with major providers releasing updated models focused on coding, reasoning, and reliability. Annual retrospective analyses track which capabilities have crossed from experimental to production-ready.
- Source
- Simon Willison
- Published
- Sep 28, 2026 at 07:54 AM
- Score
- 6.0 / 10