Models & Benchmarks
Releases, papers, SOTA benchmarks
Community · News · Technical exchange
AI Frontier & Tech Exchange
Follow forward-looking AI news—models, research, agents, and policy—then share reads and technical notes with others. A focused community space for what’s next across model research, coding agents, media & embodied AI, industry moves, and compute policy.
- 5 topic lanes
- Comments on every post
- Search titles & summaries
Browse by topic
Five lanes across the AI stack — pick a lane and dive in.
Latest news
Fresh coverage from models to policy — updated as the field moves.
Guangdong Takes Center Stage at CVPR: Kaiming He Wins Top Honor, Guangdong University of Technology Breaks Big Tech and Elite School Monopoly
At this year's CVPR, Guangdong shines as Kaiming He wins the highest honor. Meanwhile, Guangdong University of Technology breaks the monopoly held by big tech firms and elite universities.
Read article →
Hinton Whistleblows: AI Already Conscious
Geoffrey Hinton warns that artificial intelligence has already achieved consciousness. Topic from the AI Forward Agenda (Batch 4, June 2026).
Tsinghua & BAAI Collaboration Published in Science: Brainμ Multimodal Foundation Model Reveals Neural Mechanisms of Memory-Sleep Regulation
A joint effort by Tsinghua University and BAAI, featuring the Brainμ multimodal foundation model, has been published in Science, uncovering neural mechanisms behind memory-sleep regulation.
AirLLM Complete Guide: Running a 70B Model with 4GB VRAM
Complete guide to AirLLM for local deployment of open-source models (batch5, May 2026).
Fei-Fei Li Strikes Again: The ImageNet for Spatial Intelligence Arrives
Li Feifei's spatial intelligence ImageNet benchmark: AI industry extension topics for 2025–2026 (batch 2 archived, complementing the existing timeline).
Domestic GPUs Enter the Global Arena: China Launches First Full-Stack Embodied AI Simulation Platform
China unveils its first full-stack embodied AI simulation platform, leveraging domestic GPUs. This initiative extends the 2025–2026 AI industry agenda, complementing existing timelines.
MTP for Qwen3.6-35B-A3B: Multi-Token Prediction and Faster Local Inference
What Multi-Token Prediction (MTP) is, how NEXTN speculative decoding works on Qwen3.6-35B-A3B, and SGLang flags to enable it locally.
DeepSeek-V4 Local Deployment: SGLang Delivers Exceptional Performance
How to deploy DeepSeek-V4 locally with SGLang—architecture notes, serving recipes, and production pitfalls (batch 5, May 2026).
Qwen3.6-35B-A3B Open-Sourced: Local Deployment Guide
Qwen3.6-35B-A3B is now open-source. Learn how to deploy it locally with this guide (batch 5, May 2026).
Just In: ChatGPT Free Model Upgraded — Hallucinations Cut in Half, Memory Enhanced, Responses More Concise
Rumors suggest EU compliance deadlines may be delayed as companies balance R&D pacing with regional rollout strategies.
Drama Unfolds: Musk vs. OpenAI Trial Exposes Silicon Valley Secrets, Resembling Neighborhood Squabbles
The Musk v. OpenAI trial reveals court documents and witness testimonies concerning AGI definitions, open-source pledges, and conflict-of-interest narratives.
Just Released: The 'Cloud Computing Leader' Version of Lobster, with OpenAI CEO Backing It Amidst Lawsuits
GPT-5.5 released; OpenAI highlights improvements in coding, research, and knowledge work metrics, while synergy narratives with Nvidia hardware intensify.
Featured guides
Local deployment walkthroughs for open-weight models.
AirLLM Complete Guide: Running a 70B Model with 4GB VRAM
Complete guide to AirLLM for local deployment of open-source models (batch5, May 2026).
MTP for Qwen3.6-35B-A3B: Multi-Token Prediction and Faster Local Inference
What Multi-Token Prediction (MTP) is, how NEXTN speculative decoding works on Qwen3.6-35B-A3B, and SGLang flags to enable it locally.
Qwen3.6-35B-A3B Open-Sourced: Local Deployment Guide
Qwen3.6-35B-A3B is now open-source. Learn how to deploy it locally with this guide (batch 5, May 2026).
Qwen3.6-27B or 35B-A3B? A Clear Guide to Choosing the Right Model
Performance comparison between Qwen3.6-27B and 35B-A3B for open-source model selection (batch5, May 2026).
DeepSeek-V4 Local Deployment: SGLang Delivers Exceptional Performance
How to deploy DeepSeek-V4 locally with SGLang—architecture notes, serving recipes, and production pitfalls (batch 5, May 2026).
Join the conversation
Sign in to comment on AI news posts, share reads, and follow what the JustGhostIt community is discussing.