This year's most consequential interpretability work relied on open models
georgejrjrjr · x · 2026-11-19
Matt Parlmer argues that all the most consequential interpretability and safety work this year happened because researchers had access to open models — fueling the ongoing debate over whether closed-weight models shut external researchers out of inspecting and auditing model internals.
More from Models
- Karpathy unveils nanochat miniseries v1, treating LLMs as a scaling-law model family with one compute dial — eliebakouch · 2026-11-19
- First-timer quantizes DeepSeek 4.1 with ExLlamaV3's EXL3, with an AI copilot — HankYeomans · 2026-11-19
- UK Turing Institute sovereign AI model roadmap targets deployable models late 2028 onwards — bclavie · 2026-11-19
- Gemini release gaps halve each cycle: 28 to 15 to 6 days per new model — davidyin44 · 2026-11-19
- Surge AI benchmarks Gemini 4 Neon: tops Tuesday@Work at 71.6, trails on SWE — echen · 2026-11-19
- Google's rumored Neon and Carbon models reportedly match or beat Opus in early tests — almmaasoglu · 2026-11-19