Introducing GLM 5.3 on Amazon Bedrock
Amazon Web Services 1 week ago 7
Zhipu AI’s GLM 5.3 model is now available on Amazon Bedrock with fully managed APIs for coding and long-horizon agentic workloads.
Large Language Models
442 summarised stories about Large Language Models, each linking back to the original source. Browse all topics →
Key players
Monday, 5 October 2026
Meet Together Link: A Free CLI That Runs Open Models Like Kimi K3 and GLM 5.3 Inside Claude Code, Codex, and OpenCode
MarkTechPost 1 week ago 47 ● 4 sources
Together AI released Together Link, a free MIT-licensed CLI beta that plugs into existing coding agents to run open models hosted on Together AI. It installs on macOS or Linux with a single command and requires only a Together API key. The CLI adds per-session auto-routing between models like GLM 5.3/GLM 5.3 Flash (and optionally Opus 5.5) so agent users can swap models and track session costs without running a local proxy.
Reflection AI Introduces Beam: A 501B Open-Weight MoE Model With 23B Active Parameters for Coding and Agentic Workloads
MarkTechPost 1 week ago 42 ● 6 sources
Reflection AI introduced Beam, its first open-weight sparse Mixture-of-Experts model built for coding and agentic work. Beam has 501B total parameters with 23B active per token. Self-hosting is not available yet as the model is in final red-teaming, with early access via a waitlist on Reflection’s platform.
Reflection debuts Beam, an open-weight AI model to rival Chinese models at lower compute cost
TechCrunch 1 week ago 10 ● 6 sources
Reflection AI unveiled Beam, its first frontier open-weight AI model meant to match leading Chinese open models on reasoning while costing less compute. Beam has 501 billion total parameters with 23 billion active and uses a 1 million token context window. The release positions Reflection to compete with closed labs and other open models and to enable “AI factory” deployments where institutions build customized local systems using Beam’s weights and technical details.
All the drama around AI’s takeover of mathematics
The Verge 1 week ago 39 ● 19 sources
OpenAI, Anthropic, and other AI labs announced mathematics breakthroughs during the past year. The article says these results include solving one of the Millennium Prize problems. The backlash from mathematicians and academia has prompted the labs to promise they’re learning from earlier mistakes while pushing ahead.
AI Solves a Major Unsolved Math Problem. Not Everyone Is Happy
IEEE Spectrum 1 week ago 43 ● 19 sources
Leading mathematicians gathered at the Heidelberg Laureate Forum to debate how AI firms are rapidly solving major math problems, highlighted by OpenAI’s announced solution to the Navier–Stokes existence and smoothness problem. OpenAI’s announcement followed its planned 13 September forum discussions, and the article cites that in July it offered 100,000 free licenses to frontier models for academic researchers. The controversy and access concerns are shifting attitudes toward math work and evaluation, changing how researchers train, compete for jobs, and even selecting research topics away from pure verification of LLM outputs.
Yandex Introduces Sona: A Single Generative Recommender That Replaces Entire Recommendation Cascade
MarkTechPost 1 week ago 37
Yandex introduced Sona, a generative recommender model that replaces a multi-stage recommendation cascade with one served transformer combining candidate generation and ranking. In a 7-day A/B test on its smart speakers, it replaced more than 15 candidate generators plus the pre-ranking and ranking stages. The change improved metrics such as active users (+4.53%), total listening time (+6.30%), and likes (+11.42%) for the test group.
The Story of Qwen: Alibaba’s AI Models From 7B to 2.4T
MarkTechPost 1 week ago 22
Alibaba’s Qwen model line evolved from an invite-only Tongyi Qianwen enterprise chatbot beta into open-weight releases that reached 2.4 trillion parameters. The Qwen3.8 release open-sourced Qwen3.8-2.4T-A95B with 2.4 trillion parameters and 95B active parameters on August 12, 2025. As Qwen grew, Alibaba shifted from mostly restrictive licensing to broader openness for smaller models, while making larger “Plus/Max” tiers proprietary and moving some revenue-based models to a two-tier split.