Monthly logistic success-curve fits to OpenAI internal research tasks
BasilHalperin · x · 2026-11-09
BasilHalperin shares monthly logistic (METR-style) success-curve fits to OpenAI's internal research task data: one curve per month from Jan-Jul 2026, success rate vs task duration, with RMSE between 3.6-6.9 percentage points.
He says neither the logistic nor Weibull fit is clearly better, except that the METR-style approach matches the established methodology and is more comparable; what to make of the difference in results remains unclear.
More from Research
- LoRA-tuned Qwen3 skips first output line whenever the caption is under ~4 tokens — prathamdmehta · 2026-11-09
- Ramez: OpenAI's Productivity Gain Per ECI Point Appears to Decline Over Time — ramez · 2026-11-09
- Betting on transformers: incremental improvements will beat radically new LLM architectures — max_paperclips · 2026-11-09
- FragBench benchmark paper shared, details in the paper itself — arthurcolle · 2026-11-09
- Economist Paul Novosad proposes a 4-step AI-assisted peer review workflow — paulnovosad · 2026-11-09
- CoRL 2026 to host 'Do Robots Need World Models?' debate workshop in Austin — YunzhuLiYZ · 2026-11-09