Brundage: OpenAI's Spec Training May Not Be Deterministic
Miles_Brundage · x · 2026-07-07
As a continuation of previous thoughts, Miles Brundage speculated that OpenAI might not have a deterministic "Model Spec training" process internally. The related training could be somewhat arbitrary, and the personnel responsible for training might not align in their views with the Spec authors, or they might not pay enough attention to the training data. He also noted that he is not entirely certain about this.
Related event: Researcher Questions OpenAI's Model Spec Training Consistency(2 posts)→
More from Models
- Kimi K2.8 Preview rolls out: near-K3 coding performance, 1M context for all tiers — teortaxesTex · 2026-09-11
- Looking for a classifier of software engineering task shapes to pick models per task — StewartalsopIII · 2026-09-11
- DeepSeek V4 Pro API to continue after Sept 2026, billing unchanged — teortaxesTex · 2026-09-11
- DeepSeek V4.1 Flash Hits 98% of GPT-6 Astra's Score at 1.4% of the Cost in Third-Party Benchmark — ayushtweetshere · 2026-09-11
- TheZvi Polls: Has Your Coding Model Choice Changed Since Fable 5.1 and Astra? — TheZvi · 2026-09-11
- antirez Weighs In on Anthropic Banning Minors From Using Claude — antirez · 2026-09-11