Daily AI news, sourcedWhere it stands
As of Oct 2, Mercor's APEX study reported by The Decoder found Claude Opus 5.5 leading at 61.8% on the 160-task accounting benchmark, though no model fully solved nearly 60% of tasks.
- Oct 2Reported by The Decoder
Frontier models beat licensed accountants on speed and accuracy on APEX, but can't close the books alone, per The Decoder
A Mercor study reported by The Decoder had 12 licensed CPAs average about 37% on simplified APEX Accounting Benchmark tasks 18 months ago; today's best models solve those tasks almost flawlessly, faster and cheaper. On the full 160-task benchmark built with 40+ professionals, Claude Opus 5.5 leads at 61.8% of grading criteria met, ahead of Fable 5.1 at 61.0% and GPT-6 Astra at 57.9%, yet no model fully solved nearly 60% of tasks. Mercor says models still can't close the books without oversight because the study omits client communication, collaboration and accumulated context.
Source: The Decoder
- Official
- An institution announces something that has already happened, through its own channel or a founder or executive account.
- Statement by …
- A named person's views, predictions, plans or comments, including remarks reported by media when the original is unavailable.
- Reported by …
- Only media or another third party is saying it. We name the outlet.
- Unconfirmed
- Circulating online (posts, comments, leaks). No company, named person or outlet has confirmed it.
Items older than 72 hours never go in Top stories and show their original date.