Open-Source AI

Today's Stories

1. 1

   AWS ships Qwen3-TTS-12Hz-1.7B-Base voice cloning on SageMaker JumpStart

   AWS detailed deploying Qwen3-TTS-12Hz-1.7B-Base from Alibaba Cloud's Qwen team via Amazon SageMaker JumpStart to a managed real-time endpoint. It clones a voice from a few seconds of reference audio, in 10 languages, with cross-lingual cloning. Teams can now produce speech in a chosen speaker's voice, in any of 10 languages, from just a few seconds of reference audio — without retraining a model or collecting large training datasets.

   Whether deployments stay inside the roughly 24 GB GPU the walkthrough uses — the setup hinges on splitting GPU memory correctly between the talker and code2wav stages. Watch the KV cache token budget of 56,928 tokens for concurrency.

2. 2

   Intrinsic opens Intrinsic Core™ robotics platform

   Intrinsic, Google's AI robotics group, announced it is making parts of its platform open source. The package, Intrinsic Core™, includes ROS-compatible tools like Intrinsic Control, pose estimation, motion planning, grasp planning, simulation services, camera calibration, and Intrinsic-ROS drivers. Developers building robotic applications get a hardware-agnostic starting point — Intrinsic Control lets users swap robot arms, grippers, and sensors without rewriting drivers, according to the announcement. The release may lower the effort needed to assemble custom robot setups.

   Success hinges on how broadly developers adopt the open source package and the OMTS reference design — a ready-made starting point for a real-world use case. Watch the linked GitHub download for updates.

3. 3

   PrismML's 1-bit Bonsai LLM lands on Snapdragon AR1 glasses

   PrismML created a version of its tiny language models for smart glasses running on Qualcomm's Snapdragon chips. At Qualcomm's Snapdragon Summit, the chipmaker showcased PrismML's 1-bit Bonsai LLM, a 2-billion-parameter model tuned for vision and language that runs locally on AI smart glasses built on the Snapdragon AR1 Gen 1 Platform. Running the model locally on the device, rather than in the cloud, is the point of this showcase — it signals that PrismML's open-weight, on-device approach is being paired with Qualcomm's smart glasses hardware.

   No smart glasses running PrismML have been announced yet, so the showcase has not yet turned into a shipping product. Watch whether a device maker commits to the Snapdragon AR1 Gen 1 Platform version.

4. 4

   OpenCode on Amazon Bedrock: open weight coding agents

   AWS published a guide to pair OpenCode, an open-source terminal coding agent, with open weight models on Amazon Bedrock such as Moonshot AI Kimi K3, OpenAI GPT-OSS 120B, and NVIDIA Nemotron 3 Super 120B. Developers can keep code and prompts inside their own AWS account with pay-per-token billing instead of per-seat fees, and switch models by changing one API parameter, which may lower costs at scale.

   Bedrock says a global cross-Region inference profile costs approximately 10% less than a geographic profile, so the cost case hinges on whether data residency rules let teams use the global profile.

5. 5

   NVIDIA Nemotron 3 Diarization tops VoiceArena at 14.72% DER

   NVIDIA released Nemotron 3 Diarization, an open-weight 100M-parameter model that ranks #1 on VoiceArena's Diarization-Bench with a 14.72% Diarization Error Rate, supporting up to eight speakers in live and recorded conversations. The model turns overlapping multi-speaker audio into speaker-attributed transcripts, which improves search, summaries, action items, conversation analytics and voice-agent memory for meeting, call and podcast applications.

   Accuracy depends on pairing these anonymous channel labels with meeting metadata or speaker verification, and the real test is whether the eight-speaker support holds up under the customizable streaming latency.

6. 6

   Identity Digital spins out Known Systems AI

   Identity Digital spun out Known Systems AI Inc., based on its Innovation Labs unit and DNSid, a 'birth certificate for AI agents' built on Domain Name System, public key infrastructure and a blockchain ledger. Businesses already running thousands of AI agents cannot feasibly build trust with every third party one by one, so a shared accountability layer is seen as necessary for the agents to scale across platforms.

   Whether DNSid becomes a widely adopted open standard hinges on participation beyond Identity Digital, whose CEO Akram Atallah said keeping the work inside the parent 'would limit its potential.' Interim CEO Allie Kline is leading the search for a permanent leader.