AI Agents Discovered a Reasoning Strategy That Cuts LLM Tokens by 70%firethering.com·1 pts·steveharing1·0
Claude Knew It Was Being Tested. Anthropic Built a Tool to Detect Itfirethering.com·2 pts·steveharing1·0
ZAYA1-8B matches DeepSeek-R1 on math with less than 1B active parametersfirethering.com·117 pts·steveharing1·56
Ling 2.6 (Flash and 1T): Efficient Open Models Competing on Agentic Benchmarksfirethering.com·2 pts·steveharing1·0
Laguna XS.2: A Coding Model Built for Closed Environments, Now Open Weightsfirethering.com·1 pts·steveharing1·0
Xiaomi releases MiMo-v2.5 Family weights with strong coding and agent benchmarksfirethering.com·21 pts·steveharing1·7
Open CoDesign: Open-source, local-first alternative to Claude Design and v0firethering.com·11 pts·steveharing1·1
UniGenDet: Training image generators and detectors togetherivg-yanranzhang.github.io·1 pts·steveharing1·0
Prax: An agent runtime that learns from past mistakes and fixes code in a loopgithub.com·4 pts·steveharing1·0
OpenMythos: A looped transformer take on how Claude Mythos might workfirethering.com·3 pts·steveharing1·2
Anthropic's relationship with the Trump administration seems to be thawingtechcrunch.com·1 pts·steveharing1·0
Alibaba open-sources Qwen3.6-35B-A3B, a 35B MoE model with 3B active parametershuggingface.co·10 pts·steveharing1·0
Gemopus: A Gemma fine-tune that prioritizes stability over long chain-of-thoughthuggingface.co·1 pts·steveharing1·0
Bouncer: Block "crypto", "rage politics", and more from your X feed using AIgithub.com·54 pts·steveharing1·121