Open-Source AI

Today's Stories

1. 1

   NVIDIA, Hugging Face Partner on Robotics AI Tools

   NVIDIA and Hugging Face announced a collaboration on July 6 to integrate NVIDIA's physical AI capabilities into LeRobot, Hugging Face's open-source robotics library. The partnership brings NVIDIA's Isaac GR00T 1.7 reasoning model and Isaac Teleop framework to the platform. The integration makes NVIDIA's advanced robotics AI tools more accessible to developers through an open-source channel. Combining reasoning models with teleoperations (remote control) technology may lower barriers for companies and researchers building robotic systems.

   LeRobot is open-source, meaning the tools are freely available to developers. The Isaac GR00T 1.7 model and Isaac Teleop framework are now part of this shared platform.

2. 2

   AWS SageMaker HyperPod adds five enterprise inference features

   Amazon Web Services announced five new capabilities for SageMaker HyperPod inference: multi-tier data capture for auditing and model improvement, direct deployment from Hugging Face Hub, local NVMe model loading for faster cold starts, automated Route 53 DNS for custom domains, and pod-level IAM through custom service accounts. These features address practical pain points for businesses running large AI models in production — faster startup times, easier integration with popular model repositories, better security controls, and improved auditability help reduce operational friction and deployment complexity.

   The capabilities are now available in SageMaker HyperPod, AWS's service for large-scale machine learning workloads.

3. 3

   8.9 Million AI Users

   8.9 Million AI Users

4. 4

   Ollama raises $65M Series B, now used by over 8.9M developers monthly

   Ollama, an open source AI tool that helps developers run AI models on personal computers, raised $65 million（約100億円） in Series B funding led by Theory Ventures. The company has now raised $88 million（約140億円） total, bringing its monthly user base to over 8.9 million developers, with presence in 85% of the Fortune 500. Ollama addresses a real business shift toward open-weight AI models as companies seek more affordable alternatives to closed models for everyday inference tasks. Founder Jeff Morgan credits a turning point around January when larger open models became capable of complex tasks like coding, prompting enterprises and startups to view open models as viable for daily work rather than just research.

   The company operates with only 14 employees and offers cloud services on subscription tiers ranging from free to $100/month, pricing by GPU time rather than token limits. Ollama's founders previously built Docker Desktop, which established their track record in creating developer tools that reach wide adoption.

5. 5

   Databricks picks Chinese AI model GLM 5.2 for coding after matching Opus at lower cost

   Databricks tested the Chinese open-source model GLM 5.2 on its own codebase and found it tied with Anthropic's Opus 4.8 in performance while costing $1.28 per task versus $1.94 for Opus. The company now plans to make GLM 5.2 a daily workhorse for developers. Other companies—Coinbase, Lindy, and Snowflake—have similarly switched to cheaper Chinese models including GLM 5.2 and Deepseek v4. Cost pressures are shifting development workflows away from Western AI providers. On OpenRouter, Chinese models topped 30 percent of weekly traffic since February 2026, up from 11 percent last year, at 60 to 90 percent lower cost than Western alternatives. For engineering teams and smaller companies, this could mean substantial savings without sacrificing code quality—though token efficiency and routing smarter work to cheaper tiers matters as much as the model's headline price.

   Databricks found that its developers handle mostly medium-complexity tasks (61 percent), so the company plans to route more work to cheaper performance tiers based on complexity. The top-performing cluster includes Opus 4.8, GLM 5.2, and GPT 5.5 in certain configs, each hitting 82 to 90 percent pass rates; the Pareto frontier for best quality-to-cost ratio now spans models from OpenAI, Anthropic, and open-source providers.

6. 6

   ZenML launches Kitaru to help developers build production AI agents

   ZenML, an open-source MLOps (machine learning operations) platform, has released Kitaru, a new project designed to help developers build AI agents that can be deployed reliably in production. The tool addresses workflows, agent systems, fleet management, and the underlying infrastructure needed for durable, scalable deployments. Moving AI agents from experimental prototypes to working production systems requires handling challenges like reliability, observability, and the ability to replay operations — problems that traditional software deployment approaches do not fully solve. Kitaru aims to apply proven MLOps principles to this new category of AI workloads, making it easier for development teams to manage agents at scale.

   The project is part of a broader ecosystem expansion in AI tooling; ZenML points to an updated Machine Learning Tools Landscape that now includes 84 new tools, signaling rapid growth in the infrastructure category supporting AI agent development.