NEAR AI's agent solved all 672 PutnamBench math problems at a mean cost of $0.17 each, using DeepSeek V4.
Rival perfect | Hanami
NEAR AI's agent solved all 672 PutnamBench math problems at a mean cost of $0.17 each, using DeepSeek V4.
Rival perfect scores built on OpenAI's GPT-5.6 and Anthropic's Claude Opus 5 reported $44.5 and $3.52 a problem.
@NEARProtocol
Retweeted by @NEARProtocol from @MSBIntel