MarkTechPost (2 hours ago)
Block’s AI lab released Dessa, a full-duplex speech-to-speech model trained only on human conversation audio. It runs locally on a MacBook Pro M4 Max, but Block has published no size, benchmarks or weights.
Read the full story on MarkTechPost.
Google’s latest Gemini 4 Carbon AI model boasts top benchmark scores in areas like cybersecurity and coding.
One AI “super user” in a team of five to seven people lifts the whole team’s AI adoption and usage by 56%, Atlassian’s Avani Prabhakar told TNW. Atlassian counts someone as a super user if they use AI’s most sophisticated features more than 40 times a week to change a workflow, she said.
Kimi K3.1 Pro is no longer the fourth best AI model in the world.
Datadog has released Scout, a model trained to investigate production problems. It runs inside Bits Investigation, the company’s AI SRE product, where it can pick up an alert, look through telemetry and suggest what caused the problem.
Chinese AI companies DeepSeek and Moonshot AI relied heavily on OpenAI’s GPT-6 Astra to train their latest models, The Information reported on Wednesday, citing people with direct knowledge of the matter.
Infini-AI-Lab released Hermit, an Apache-2.0 harness adapter framework that trains an external layer of instructions, skills, memory and tools from task feedback, leaving the agent’s model and harness unchanged.
AssemblyAI has released AssemblyAI-TTS-1, its first text-to-speech model for voice agents. It costs $15 per 1M characters, offers unlimited concurrency without a contract, and supports 6 languages at launch.
NVIDIA released Nemotron 3.5 Transcribe, an open-weights speech LLM with 4.3B active parameters. It transcribes 40 languages, adds speaker labels and word timestamps, and posts a 4.71% mean WER on the English Open ASR Leaderboard.
Google DeepMind introduced Gemini 4 Carbon, a frontier model that scores 59 on the Artificial Analysis Intelligence Index and 88.2% on SEC-bench Pro. Access starts with vetted cyber defenders, at $2/$10 per million tokens.
MiniMax has open-sourced MCode Stack, an MIT-licensed repo of 17 SKILL.md skills and 5 checklists that its MiniMax Code team uses daily for code review, subagent splits, context budgets and long-running commands.
71 active sources
26 Turkish · 45 English
All sources
