The company said that the model was trained on 15 trillion mixed visual and text tokens.

On SWE-Bench Verified, the model achieved a score of 70.6%. This performance is notably competitive when placed alongside ...

Kimi K2.5 handles up to 100 sub-agents and 1,500 tool calls, cutting task time 4.5x so you finish complex work sooner.

The b3 is built around a new idea called threat snapshots. Instead of simulating an entire AI agent from start to finish, threat snapshots zoom in on the critical points where vulnerabilities in large ...

The UK AI Security Institute (AISI) has partnered with the commercial security sector on a new open source framework designed to help large language model (LLM) developers improve security posture.

Kimi has a standard mode and a Thinking mode that offers higher output quality. Additionally, a capability called K2.5 Agent ...

Abu Dhabi-based Mohamed bin Zayed University of Artificial Intelligence’s (MBZUAI) Institute of Foundation Models has released K2 Think V2, a 70 billion-parameter open-source reasoning model that the ...

30-person startup Arcee AI has released a 400B model called Trinity, which it says is one of the biggest open source foundation models from a US company.

Mistral AI has launched Voxtral Transcribe 2, a new on-device speech-to-text model family featuring real-time transcription, ...

OpenAI has released two new open-weight language models under the permissive Apache 2.0 license. These models are designed to deliver strong real-world performance while running on consumer hardware, ...