xAI Ships Grok 4.7 With Longer Agent Runs at Unchanged Prices
xAI ships Grok 4.7 with longer deliberation, better self-checking, and stronger safety guardrails, matching Grok 4.6 on price and latency.
// Intelligence
Llms
A focused feed of models, agents, research and open-source releases for people building with AI.
Updated
Xiaomi's MiMo-V2.6-Flash-RL Opens a 309B Agent Model Trained for $850K
Xiaomi's new omnimodal Mixture-of-Experts model activates 15B parameters, ships under MIT license, and pushes reinforcement learning into a self-improvement loop.
Xiaomi's MiMo-V2.6-Pro Tops 114 Open Models With a 1T Parameter Beast
Xiaomi's new omnimodal MoE debuts as the top open-weights model on Artificial Analysis, trained with a single mixed RL run for around $2.6M.
Microsoft's RetroChimera Plans Drug Synthesis Routes and Wins 9 of 10 Expert Reviews
Microsoft Research's RetroChimera combines two neural networks to plan chemical syntheses that expert chemists prefer over published literature reactions.
Jared Palmer's Open-Source Kev Answers Typed Questions in one Pass
Jared Palmer's open-source Kev family scales a Jev-style decision model up to 8B parameters using Qwen3, LoRA, and a pointer head.
IsValorum Squeezes a 35B Open-Source Qwen3.8 Model Into 14.65 GB
A 35B Qwen mixture-of-experts release strips refusal vectors and squeezes the whole model plus 256K context into 24GB of VRAM.
Open-Source Laya Beats Closed Jev API at 7x Faster Decision Speed
Laya is an open-source, non-autoregressive decision engine that answers typed questions across 100+ languages in a single 33ms forward pass.
Qwen3.8 Squeezed to 11.8 GB Runs at 150 Tokens per Second
A community remix packs an uncensored 27B Qwen into 11.8 GB by giving each of 851 tensors its own quantization type, hitting 100+ tokens per second on a 16 GB card.
HyperQwen Runs Qwen3.8-27B on a Single RTX 3090 at 1,035 tok/s
HyperQwen crams a 27B Qwen model onto one 24GB RTX 3090 with vLLM patches, hitting 127 tok/s solo and ~1,035 tok/s at 64 concurrent.
Alibaba's Qwen3.8 LiveTranslate Cuts Speech Translation Lag to 2.3 Seconds
Alibaba's next-gen simultaneous interpretation model cuts average lag to 2.3 seconds across 60 languages, adds speaker diarization, and clones each voice separately.
Researchers Found a 'Pain Axis' That Makes Qwen 2.5 Pay for Relief
Researchers isolated a linear direction in 25 open-weight LLMs that behaves like pain, distinct from fear or sadness, and drives self-relief behavior.
OpenJev Runs Agent Decisions 5x Faster by Skipping Token Generation
OpenJev reproduces TypeSafe's semantic if-statement service using open 4B models on a single RTX 3090, reading option logits directly instead of parsing generated JSON.
Hikari07jp Edits Ternary Bonsai 2 to Drop Refusals From 37 to 0
A community build strips refusals from a 27B ternary model by rewriting 2-bit codes in place, with no dequantization and zero size change.
Convai Innovations' Laya Beats Closed Decision Models at 32.8 ms
Convai Innovations released Laya, an open-source non-autoregressive decision model that returns typed answers with calibrated probabilities in ~33ms across 100+ languages.
Shanghai AI Lab Releases Atria Dawn Preview to Beat GPT on Research
Shanghai AI Lab open-sourced Atria Dawn Preview, a 744B-parameter Mixture-of-Experts agentic model with MIT-licensed weights and FP8 checkpoints.
Sakana AI Bets Against Transformers With Its Frontier Intelligence Group
Sakana AI launches the Frontier Intelligence Group, a research collective betting that Transformers and scaling laws aren't the final answer to AGI.
OpenAI's Astra for Law Beats Web Search by 40% on Legal Research
OpenAI wraps GPT-6 Astra in a legal search index of 230 million URLs, hits 54% correctness on Vals AI's Legal Research Bench.
Epoch Audits 15 AI Benchmarks and Flags Nine as Flawed
Epoch AI launched Benchmark Reviews, auditing 15 popular AI benchmarks with only 4 passing as Verified and 9 flagged as Flawed.
PrismML Squeezes Qwen3.8 27B Into 5.9 GB With 98% Performance Retained
PrismML's ternary-quantized 27B model retains 98.2% of full-precision Qwen3.8 27B performance in a 5.9GB footprint, hitting 143 tokens/sec on an RTX 5090.
Anthropic Reveals Claude Now Leads 26% of Its Own AI Research
Anthropic published three transparency metrics tracking AI-driven R&D, agent oversight, and compute allocation, urging other frontier labs to adopt the same reporting.
OpenAI's Astra for Law Pushes Into Big Firms With 54% Research Accuracy
OpenAI unveils a legal-specific configuration of GPT-6 Astra with a 230M-URL search index, firm-built workflows, and 73 plugins for practitioners.
Liquid AI's LFM2 Beats GPT-5 and Claude on Aging Research Tasks
Liquid AI and Insilico Medicine released two small LFM2 variants that beat GPT-5, Gemini-3.1-Pro, and Claude Opus on aging biology benchmarks.
GLM-4.7-Flash Roleplay Fine-Tune Lands Calibrated GGUF Builds for Local Inference
A community imatrix-quantized GGUF build of an uncensored GLM-4.7-Flash fine-tune brings local Chinese and English roleplay to consumer hardware.
Sakana AI's Fugu Max Beats Frontier Models at 60% Lower Cost
Sakana Chat now runs on the Fugu Max orchestrator model and gains persistent memory across conversations, closing the gap with ChatGPT-style assistants.
No stories match these filters.