LEANN is a low-storage approximate nearest neighbor (ANN) index built on top of FAISS. It targets one problem: vector indexes ...
Liquid AI LFM3 ships 6 open-weight 2B-class models; LFM3-1.7B averages 49.37 across 29 self-reported benchmarks, above ...
FlashMap speculative decoding uses a 0.5B flow map drafter to reach 6.14x lossless speedup on Qwen3-8B, beating DFlash.
Google DeepMind’s Genie 4 world model generates interactive worlds with native audio, dialogue, and Gemini 3.8 Flash planning.
OpenAI’s GPT-6.1 Terra brings fast computer use at $1/$5 per million tokens, scoring 60.0% on Agents’ Last Exam.
Mirendil Fern 0.1 is a from-scratch coding model with a 10M-token context window, built for agentic AI research.
Nace AI’s Drex for Search reports 92.05% on BrowseComp-Plus with GPT-5, using only 12.6 search calls per question.
Baidu open-sources Unlimited OCR 2, an MIT-licensed 3B OCR model scoring 77.1 on MarathonDocBench for hundred-page document parsing.
Performance: 66.7% average on 5 DNA benchmarks, best of every open model tested (self-reported). Best: 74.9% on BioReason ...
Microsoft open-sourced Agensh, a decentralized multi-agent harness that scales to 1,024 agents, lifting Pandoc pass rates to ...
TokenGate adds learned token pruning to existing LLMs, cutting long-context prefill time up to 11.3x and memory 3.4x.
The research report shows a Reachy Mini robot running on a Jetson Orin Nano 8GB. In that demo, Cosmos 3 Edge handles vision ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results