GPT-4.1
IBM's ALTK-Evolve lifts GPT-4.1 to the top of τ²-bench retail
IBM Research says its ALTK-Evolve method took a basic customer-service agent built on GPT-4.1, an older non-reasoning model, to 67.5% Pass^4 on τ²-bench retail.…
Daily Releases PhoneLLM Alpha 2 With Two Open-Weights Voice Agent Models
Daily has released Alpha 2 of Pipecat PhoneLLM, an open-weights text model built for voice agent pipelines, now in two variants: PhoneLLM Mini and the larger PhoneLLM…