Prime Intellect
Pydantic benchmark: code mode roughly halves cost on long agent tasks
Letting a model write a short program instead of calling tools one at a time cut cost by 58% and wall time by 51% on long tasks, according to a benchmark published by…
Goodfire's probe-based cyber monitors beat an LLM judge at 50x lower cost
Goodfire Research says it has built cybersecurity monitors for Kimi K3 and GLM 5.3 that outperform an LLM judge at about 50x lower cost, even when run live on every…