Benchmarks - Page 23
Stories from Benchmarks come first, newest first, then stories from connected topics (AI Models, LLM, AI, coding agents, preprint, open weights), each labelled with where it comes from. This view is not indexed by search engines.
Latest
UltraBERT Explained: Open 32K-Context Encoders, and Why “How Much You Hide” Decides What a Model Is Good At
DMIS Lab's UltraBERT is a family of open 32K-context encoders. Explainer: what encoders do, and why the masking finding matters for search and RAG.
Inworld Launches Inworld Inference, an API It Says Is Built for Consumer AI Apps
Inworld has opened Inworld Inference, an API for running open AI models in consumer apps. It says it is faster and cheaper than the median provider.
Chainguard Launches Three Tools to Keep AI Coding Agents Contained, Checked and Hunting for Bugs
Chainguard launched three tools for AI coding agents: Workspaces and Checks in beta, plus Harness, which uses Claude Mythos 5.1 to find bugs.
Gemini 4 Argon Is Now Generally Available in GitHub Copilot: Who Gets It, What It Costs and What Admins Need to Do
GitHub has made Google's Gemini 4 Argon generally available in Copilot for Pro+, Max, Business and Enterprise users, billed at provider list pricing.
Cursor Opens Origin App Marketplace With More Than 20 Partner Apps
Cursor's Origin code host now has an app marketplace with 20+ partner apps, from Cloudflare to Linear, and is open to public submissions from today.
Liquid AI Releases LFM2.5-24B-A2B, an Open-Weight Model for Running AI Agents on Your Own Device
Liquid AI's open-weight LFM2.5-24B-A2B has 24B parameters but uses only 2B at a time, built to run AI agents fast and privately on your own device.
AMD Launches Its Next Ryzen AI Halo Developer Platform, With Up to 192GB of Memory for Running Large AI Models Locally
AMD has launched its next Ryzen AI Halo developer platform, with the Ryzen AI Max+ PRO 495 and up to 192GB of memory for running big AI models locally.
AWS Makes Its AI/ML Skill a Core Part of the Agent Toolkit, Taking Coding Agents From Inference Tuning to Full SageMaker Workflows
AWS's AI/ML skill is now a core Agent Toolkit skill, so coding agents can select, train, optimise and deploy SageMaker AI models in plain language.
Google’s Agent Development Kit Now Covers Five Languages: Python, Go, TypeScript, Java and Kotlin
Google's Agent Development Kit (ADK) now covers Python, Go, TypeScript, Java and Kotlin, with MCP and A2A support in every language, Google says.
LightOn Releases ColGrep-Agent-2B, an Open Local Code-Search Agent That It Says Cuts Claude Code’s Tokens by a Third
LightOn's open 2B-parameter ColGrep-Agent searches code locally for coding agents; it says Claude Code used 34% fewer API tokens than with grep.
Latest on TSN
Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.
