Z.ai released GLM-5.3-Flash, a multimodal model with 320 billion parameters and 18 billion active parameters, supporting coding and agentic workloads with contexts up to one million tokens.
Perplexity's Portable Computer runs its orchestrator, subagent, and tool harness locally on NVIDIA DGX Spark hardware, keeping files on-device and making cloud escalation opt-in per step. It supports PPLX 27B or Qwen 3.8 27B and is available on Linux to Pro and Max subscribers.
OpenAI published initial benchmarks for Jalapeño, an inference chip developed with Broadcom, claiming speed and efficiency advantages over Nvidia GB200 and GB300 systems. The chip is designed for inference rather than training and is expected to roll out through 2027.
Figure launched a robot-training dataset containing 16 million real-world videos collected across 108 countries.
Claude Security can use Mythos 5 to scan enterprise codebases, identify vulnerabilities, and suggest patches for human review. Anthropic is also working with security vendors and committed $35 million in Claude credits to help open-source projects address vulnerabilities.
Sign in to read the full issue and every issue in the archive. Free for 14 days.
Anything about the issue as a whole — a story that should have been here, a category that is off, two entries that are really one event, or how it reads.