Evaluator Bench
Epoch AI (Substack), 2026-03-05; retrieved 2026-09-15 tier 3 self (epoch)confirmed

GPT-5.4 set a new record on FrontierMath

https://epochai.substack.com/p/gpt-54-set-a-new-record-on-frontiermath

States 'We had pre-release access to evaluate the model' and that FrontierMath was funded by OpenAI. Fetched 2026-09-15; span present.

Signals citing this source

  • Epoch AI, A (for): Epoch reports pre-release access to evaluate GPT-5.4 (March 2026) and Meta's Muse Spark on FrontierMath.“We had pre-release access to evaluate the model”

Ledger rows citing this URL

  • none