unknown

GPT-5.4 and GPT-5.4 Pro from @OpenAI on ARC-AGI Semi Private

ARC-AGI-2:
- GPT-5.4: 74.0%, $1.52/task
- GPT-5.4 Pro: 83.3%, $16.41/task https://t.co/vpBlCIDrUb

TestingCatalog News 🗞

GPT-5.4 has scored 74.0% on the ARC-AGI-2 benchmark, with GPT-5.4 Pro reaching 83.3%, approaching the performance level of Gemini 3 Deep Think on this reasoning evaluation.

GPT-5.4 scored 74.0% on ARC-AGI-2 👀

GPT-5.4 Pro got 83.3%, getting close to a level of Gemini 3 Deep Think. https://t.co/hdwWb8XCyx https://t.co/7YkgLY0zzk

<p>GPT-5.4 scored 74.0% on ARC-AGI-2 👀

GPT-5.4 Pro got 83.3%, getting close to a level of Gemini 3 Deep Think. https://t.co/hdwWb8XCyx https://t.co/7YkgLY0zzk</p>