Google's Sleep Framework Lets AI Models Keep Learning Without Forgetting
Google Research proposes a biologically inspired sleep phase that lets LLMs consolidate new facts into weights without forgetting, scaling context to 10M tokens.
// Intelligence
Llms
A focused feed of models, agents, research and open-source releases for people building with AI.
Updated
Liquid AI's LFM3 Shrinks Six Specialized Models to Run on Your Phone
Liquid AI's LFM3 family packs text, vision, audio, omni, and decision models into one 2B backbone built for on-device fine-tuning.
Google DeepMind's Genie 4 Builds Interactive Worlds With a Gemini Planner
Google DeepMind's new world model streams photorealistic, playable 3D environments with native audio, live physics, and Gemini-driven reasoning in real time.
DeepVK's USER2 Beats 7B Models at Russian Search With Just 149M Parameters
DeepVK's new Russian sentence encoder packs 8,192-token context and Matryoshka embeddings into a 149M-parameter model that rivals systems 4x its size.
Parallel Web Systems' Decisions API Answers 100 Questions for $0.001
Parallel's new Decisions API combines fast classification with live web retrieval, returning typed yes/no answers with citations at $0.001 per call.
Physicists Train AI to Discover Its Own Order Parameters From Spin Data
A new paper treats atoms as agents chatting through Bayesian inference, letting a neural model discover order parameters from raw spin snapshots alone.
Artificial Analysis' Code2Video-AA Benchmark Shows Claude Opus 5.5 Dominates AI Video Coding
A new benchmark judges 27 models on turning marketing briefs into animated HTML videos, with Claude Opus 5.5 leading at 10x the cost of rivals.
Cursor Adds DeepSeek V4.1 Flash Hitting 47.7% for Under $1 per Task
DeepSeek V4.1 Flash lands in Cursor with four effort tiers that sit on the CursorBench cost frontier from $0.60 to $0.98 per task.
OpenAI's GPT-6.1 Terra Cuts Agent Costs 50% and Runs 3x Faster
OpenAI's new middle-tier model matches GPT-6 Astra on agentic benchmarks at one-tenth the price and triple the speed.
OTel 2.0 Turns Google's Gemma 4 Into a One-Token Decision Model
A 4-bit Gemma 4 MoE build turns a telecom LLM into a probability-driven decision engine, running at 32k tokens per second on a single AMD MI350P.
NVIDIA's Nemotron 3.5 Super VL Hits 22.7% on ARC-AGI-1 at $0.03 per Task
ARC Prize ran NVIDIA's first open vision-language flagship through its reasoning gauntlet, and the numbers reveal where hybrid Mamba-MoE models still trip.
Sakana AI Plans Sawara, a Japanese Foundation Model Built From Scratch
Sakana AI unveils Sawara, a from-scratch Japanese foundation model, alongside a three-pillar strategy spanning pre-training, post-training, and orchestration.
Anthropic Files for IPO Revealing $16B Revenue and $80B Loss
The Claude maker is heading to Nasdaq under ticker ANT with $16.3B in half-year revenue, an $80.7B loss, and a potential $2T valuation.
Lyra C's Neuron Explorer Lets Researchers Steer Qwen3.5-4B's Hidden Channels
A new open-source tool pushes every MLP neuron in Qwen3-4B up and down, saves every rollout, then lets you browse the model as a 3D map.
One Steering Vector Now Controls Both Image and Text Generation
A new preprint shows that steering vectors extracted from text can control image generation and vice versa inside unified multimodal models, zero-shot.
Cornell's DecisionRAG Beats Dense Retrieval by 17 Points Without Training
DecisionRAG replaces vector search with an LLM that judges candidates in parallel, beating dense retrieval plus an 8B reranker on multiple benchmarks.
NVIDIA Releases Nemotron 3.5 Lightning With 1M-Token Context on One GPU
NVIDIA's new 30B hybrid Mamba-Transformer MoE activates just 3B parameters, runs on a single H100, and handles up to 1M tokens of context.
DeepReinforce's Ornith 1.0 Drops Uncensored GGUF Builds for Local Coding
A new batch of GGUF quantizations makes an uncensored, abliterated Ornith 1.0 9B coding model runnable on an 8GB laptop through llama.cpp or Ollama.
Huihui-ai Brings DeepSeek V4.1 Flash's 763B Parameters to Local Machines
Huihui-ai ships uncensored GGUF builds of DeepSeek V4.1 Flash 1017, including sub-2-bit Strata quantizations that make a 763B MoE runnable from an SSD.
ISTA-DASLab Squeezes DeepSeek's 763B Model onto a 12 GB GPU
ISTA-DASLab shrinks DeepSeek-V4.1-Flash-1017 to under 2 bits per weight, letting a 763B MoE run locally on a single 12GB consumer GPU.
Datawhale Rebuilds Stanford's CS336 Into a Chinese Open-Source LLM Course
A Chinese-language adaptation of Stanford's CS336 bundles 16 chapters and 6 hands-on assignments covering the full LLM stack, from BPE to GRPO.
Community-Built PRISM Merges MiniMax-H3 Quality With Eight-Step Speed
A community-published merge for MiniMax-H3's Ref2VA video model uses SVD denoising and geodesic interpolation to fuse a quality finetune with an 8-step turbo variant.
OrcaRouter's OrcaCyber Nano 1.0 Lets Security Teams Analyze Malware Locally
OrcaRouter dropped a 27B vision-language model tuned for reverse engineering, vuln analysis, and secure code review, packaged as GGUF to run fully offline.
Teaching Small AI Models to Learn From Their Own Mistakes
A new literature review unifies the fragmented field of on-policy distillation, giving the student-rollout training loop a shared f-divergence framework and three-axis taxonomy.
No stories match these filters.