Can VLMs replicate publication-ready figures but without agent?. Many studies like this one. PlotGen: Multi-Agent LLM-based Scientific Data Visualization via Multimodal Feedback. shows that complex multi-agent, multi-modal stuffs can do this. But in

RT @scaling01: GPT-OSS ranking 34th on SimpleBench

Had a vibe test with gpt-oss-120b on Drylab AI platform to analyze biomedical data. Plus: it's fast, perfect tool calling .Minus: failed to generate runnable code .It's worse than kimi-k2 on this dataset and system prompt. Might need more prompt tuning?.

RT @DrylabAI: In most papers: “We ran this code. Trust us.” . In Drylab: “Here’s the code AI wrote. Here’s the environment. Test it, tweak….

RT @DrylabAI: We asked Drylab to help design protocols. It said: “Sure, I’ve read 10,000 papers.” Now it’s suggesting steps, optimizing wo….

RT @jaseweston: 🤖Introducing: CoT-Self-Instruct 🤖.📝: - Builds high-quality synthetic data via reasoning CoT + quali….

RT @iScienceLuvr: Rubrics as Rewards: Reinforcement Learning Beyond Verifiable Domains. 'We introduce Rubrics as Rewards (RaR), a framework….

RT @alexwei_: 1/N I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI….

RT @DrylabAI: One of our shining use cases is scRNA-seq, with an average runtime of ~7 minutes. 🧬3 questions we usually receive from our e….

RT @BiologyAIDaily: BioMedTools: a language model-powered community for biomedical computational tools. １．BioMedTools introduces a novel AI….

RT @benln: For any pre-seed founders exploring accelerators, this is my shortlist:. • @HF0Residency (SF).• @southpkcommons (SF/NYC/Bangalor….

At Drylab AI, we fine-tuned the best AI Agent for bioinformatics. Stay tuned for the open source version in next weeks !.

RT @DrylabAI: 7 days of analysis. Now done in 7 minutes. AI for biologists is here .🧪🧬🧫🔬👩🏻‍🔬🧑🏾‍🔬🧑‍🔬👩🏼‍🔬. #scRNAse….

RT @Grad62304977: Seems like this just worked (better?).

RT @Agentica_: 🚀 Introducing DeepSWE 🤖: our fully open-sourced, SOTA software engineering agent trained purely with RL on top of Qwen3-32B.….

RT @rohanpaul_ai: PDF parsing is still painful because LLMs reorder text in complex layouts, break tables across pages, and fail on graphs….

RT @StasBekman: My first project at @Snowflake AI Research is complete! . I present to you Arctic Long Sequence Training (ALST) . Paper: ht….

RT @Dahoas1: Really interesting long-context agentic planning details in the Gemini-2.5 tech report:

RT @hwchase17: ❓Why do some AI products explode in adoption while others struggle?. It's not just related to model capabilities - there's a….