AllLLMRoboticsStartupsResearchEthicsBusinessHardwareAutonomousMedicineCreative
Topic hub

Open-Source AI

Open models — Llama, Mistral, DeepSeek, Qwen and others — are catching up to closed labs and let you run AI on your own hardware, vendor-free. That shifts the balance of power, the safety debate and the economics of the field. Here is everything on open-source AI: open-weight releases, licenses, local deployment and the race against closed models.

Official site ↗1225 articles
In the glossary:Open-Source AIOpen-Weights ModelLarge Language Model (LLM)Foundation Model
Anthropic Launches Free OSS Scanner Vulnerability Scanner for Open-Source Code
LLM·Anthropic Research

Anthropic Launches Free OSS Scanner Vulnerability Scanner for Open-Source Code

Oct 24, 2026·3 min
Qwen-Image-PE Teaches Neural Networks Design: Prompt Enhancer Adds +7.81 Points
LLM

Qwen-Image-PE Teaches Neural Networks Design: Prompt Enhancer Adds +7.81 Points

Oct 24, 2026·2 min
Open-Source Vision Backbone Learns to Change Its Own Learning Rule on the Fly: 1000+ GitHub Stars
LLM

Open-Source Vision Backbone Learns to Change Its Own Learning Rule on the Fly: 1000+ GitHub Stars

Oct 23, 2026·2 min
Zhejiang University and Peking University Propose SPADER for Multi-Answer QA
LLM

Zhejiang University and Peking University Propose SPADER for Multi-Answer QA

Oct 23, 2026·2 min
Baidu, Tsinghua and Renmin Open-Source LoopVL, Teaching a Loop Transformer to See
LLM·Jiqizhixin (机器之心)

Baidu, Tsinghua and Renmin Open-Source LoopVL, Teaching a Loop Transformer to See

Baidu, Tsinghua University and Renmin University's Gaoling School have open-sourced LoopVL, a vision-language model built on a Loop Transformer with shared weights.

Oct 23, 2026·2 min
Vercel Launches Python in General Availability: Six SDKs and FastAPI Support
LLM·Vercel Blog

Vercel Launches Python in General Availability: Six SDKs and FastAPI Support

Vercel launched Python in general availability, enabling developers to deploy Python applications alongside JavaScript frontends with automatic framework detection…

Oct 23, 2026·2 min
DeepSeek credits users with tokens as compensation after multi-hour outage
LLM·Jiqizhixin (机器之心)

DeepSeek credits users with tokens as compensation after multi-hour outage

On the morning of October 15, DeepSeek's web chat and API, including V4 Pro and V4.1 Flash, were down for several hours. The company made a rare move for the…

Oct 23, 2026·2 min
KTransformers Moves Part of the MoE Computation to the CPU and Eliminates Weight Transfers During Fine-Tuning
LLM·Jiqizhixin (机器之心)

KTransformers Moves Part of the MoE Computation to the CPU and Eliminates Weight Transfers During Fine-Tuning

The KTransformers team proposes running MoE experts directly on the CPU during fine-tuning, aiming to tune open-source giants like DeepSeek-V4 and Kimi-K3 without a…

Oct 23, 2026·2 min
Top themes
01ProD Stops AI Agents From Looping: Tested on 8 Models (0.6B–8B)Oct 23, 2026·2 min02Groq Launches Meta Muse Spark Open: 816B Parameters and 350+ Tokens/sOct 23, 2026·2 min03VAST Open-Sources a World Model: Anime Version of Flash Step Runs in Real TimeOct 23, 2026·2 min04RSIHub Packs 4 Agent Evolution Methods Into One Framework: Up to +13.5 ppOct 23, 2026·2 min05Hongbo Sun Proposes an Algorithm for Variational Inequalities: 6400 Problems, 1e-5Oct 23, 2026·2 min
Tsinghua's LoopWAM Tops RoboChallenge Table30 and Goes Open Source
LLM·Jiqizhixin (机器之心)

Tsinghua's LoopWAM Tops RoboChallenge Table30 and Goes Open Source

Oct 22, 2026·2 min
Cohere to acquire AISLE, a startup that found 500+ CVE-assigned vulnerabilities in its first year
LLM·Cohere Blog

Cohere to acquire AISLE, a startup that found 500+ CVE-assigned vulnerabilities in its first year

Oct 22, 2026·2 min
LEASH hands hard steps to a strong agent and cuts Opus 5 costs by 19.2%
LLM·Hugging Face Daily Papers

LEASH hands hard steps to a strong agent and cuts Opus 5 costs by 19.2%

Oct 22, 2026·2 min
ACT3D teaches a VLM to explore 3D scenes and beats Gemini 3.6 Flash: 53.2% across six benchmarks
LLM·Hugging Face Daily Papers

ACT3D teaches a VLM to explore 3D scenes and beats Gemini 3.6 Flash: 53.2% across six benchmarks

Oct 22, 2026·2 min
Mistral releases Timestral: a 209M-parameter model that tops three benchmarks
LLM·Mistral AI News

Mistral releases Timestral: a 209M-parameter model that tops three benchmarks

Mistral's first time-series model family ranks first on GIFT-Eval, TIME and fev-bench with 209M parameters, and it runs on an ordinary server CPU.

Oct 22, 2026·2 min
DeepSeek open-sources its DeepInfer inference engine, and analysts calculate an 82% margin
LLM·Jiqizhixin (机器之心)

DeepSeek open-sources its DeepInfer inference engine, and analysts calculate an 82% margin

DeepSeek has released its internal DeepInfer framework and the deepseek-recipe component under Apache 2.0. Third-party estimates put the gross margin of its AI…

Oct 22, 2026·3 min
Competence Over Accuracy: Why GPT and Qwen Lose Logic When Problems Get Rephrased
LLM·Hugging Face Daily Papers

Competence Over Accuracy: Why GPT and Qwen Lose Logic When Problems Get Rephrased

A new methodology called Competence-Based Evaluation (CBE) has uncovered a fundamental flaw in LLM logical reasoning: even strong GPT and Qwen models fail on the same task type when its wording change

Oct 22, 2026·2 min
MOSS-Image: Open 5B Model Renders Long Text with 96.3% Coverage
LLM·Hugging Face Daily Papers

MOSS-Image: Open 5B Model Renders Long Text with 96.3% Coverage

Oct 22, 2026·2 min
DeepSeek releases V4.1-Flash-1017 and opens its Search API at $0.75 per 1000 calls
LLM·DeepSeek Blog

DeepSeek releases V4.1-Flash-1017 and opens its Search API at $0.75 per 1000 calls

Oct 22, 2026·2 min
TokenWorth Teaches AI Agents to Forget: Best Results on Three Environments With the Fewest Tokens
LLM·Hugging Face Daily Papers

TokenWorth Teaches AI Agents to Forget: Best Results on Three Environments With the Fewest Tokens

Oct 22, 2026·2 min
BrickWright Teaches Coding Agents to Build with LEGO: 100% Valid Builds vs. 9–69%
LLM·Hugging Face Daily Papers

BrickWright Teaches Coding Agents to Build with LEGO: 100% Valid Builds vs. 9–69%

Oct 22, 2026·2 min
Cola2 beats Llama-3-8B on HellaSwag with 43 times fewer tokens
LLM·Hugging Face Daily Papers

Cola2 beats Llama-3-8B on HellaSwag with 43 times fewer tokens

Oct 22, 2026·3 min
Claude Proves Safety of 123 Libraries at the RISC-V Machine-Code Level
LLM

Claude Proves Safety of 123 Libraries at the RISC-V Machine-Code Level

Oct 22, 2026·3 min
WeChat Open-Sources WeVisOne-VL: Detection, Masks and Matting in a Single VLM
LLM

WeChat Open-Sources WeVisOne-VL: Detection, Masks and Matting in a Single VLM

Oct 22, 2026·3 min
Anthropic opens a call for life sciences researchers: 8–12 weeks with Claude, free of charge
LLM

Anthropic opens a call for life sciences researchers: 8–12 weeks with Claude, free of charge

Oct 22, 2026·3 min
Cognition releases agent plugin to fix PHP and Node.js holes, including a CVSS 9.2 flaw
LLM·Windsurf Blog

Cognition releases agent plugin to fix PHP and Node.js holes, including a CVSS 9.2 flaw

After the critical PHP2Shell flaws and a Node.js release fixing 10 CVEs, Cognition open-sourced a plugin that has Devin and other coding agents find and patch…

Oct 22, 2026·3 min
Echo Teaches VLA Robots to Recognize Success and Failure Inside Their Own Actions
LLM·Hugging Face Daily Papers

Echo Teaches VLA Robots to Recognize Success and Failure Inside Their Own Actions

Researchers embedded stop and failure signals in every action chunk of a VLA model and denoise them faster than the motions. No per-frame progress labels are needed.

Oct 22, 2026·2 min