🔴 High Significance
Model Releases
🔴 We should heavily discourage and moderate cloud API (deepseek api, GLM api, etc.) topics and discussion. This is LOCAL first. — score 75
Sources: reddit/r/LocalLLaMA
I’m just some fucking guy. This is just some fucking opinion. I’ve seen tons of stealth marketing or related topics on this subreddit about how great or how easy it is to use some random subscription api. Why the fuck are we allowing people to so casually talk about how much more affordable their za
Developer Tools
🔴 shuvonsec/claude-bug-bounty — AI-powered bug bounty hunting from your terminal - recon, 20 vuln classes, autonomous hunting, and report generation. All inside Claude Code. — score 86
Sources: github_trending
AI-powered bug bounty hunting from your terminal - recon, 20 vuln classes, autonomous hunting, and report generation. All inside Claude Code.
🔴 browser-use/browser-use — 🌐 Make websites accessible for AI agents. Automate tasks online with ease. — score 81
Sources: github_trending
🌐 Make websites accessible for AI agents. Automate tasks online with ease.
🔴 How to setup a local coding agent on macOS — score 75
Sources: hackernews
Business & Funding
🔴 Anthropic forced to abruptly disable Fable 5 & Mythos 5 globally by US Gov over a jailbreak. This is exactly why we need local models. — score 89
Sources: reddit/r/LocalLLaMA
I just saw this statement regarding Anthropic being hit with an emergency export control directive from the US government. They were forced to pull the plug on Fable 5 and Mythos 5 for all customers globally. The tl;dr is that the government got spooked by a narrow jailbreak (which basically just so
Other Signals
🔴 Friendly reminder — score 96
Sources: reddit/r/LocalLLaMA
If you don't have it on your own drive, someone is going to take it away, enshittify it, bar you from accessing it, censor it, and hike the prices of it sooner or later.
🔴 Open source AI must win — score 92
Sources: hackernews
🔴 MiniMaxAI/MiniMax-M3 · Hugging Face — score 82
Sources: reddit/r/LocalLLaMA
Minimax m3 weights are out !! It has ~428B parameters and ~23B activated parameters.
🔴 Unprofessional Coauthor Behavior with Hallucinated References [D] — score 81
Sources: reddit/r/MachineLearning
Just thought I'd highlight this issue to the ML community, since I recently had this problem arise and it might be useful for some. I had a coauthor who I knew was somewhat untrustworthy when it came to LLM use. This coauthor added some last-minute new references to the paper. The deadline was near,
🟡 Notable
Model Releases
🟡 We should set up a torrent network for open source models. — score 54
Sources: reddit/r/LocalLLaMA
Was just thinking about this due to recent events. Hugging Face is a US-based company, legally incorporated as Hugging Face, Inc. with its official headquarters located in Brooklyn, New York. It seems like a pretty big single point of failure for local models. Maybe a distributed network mirror of m
🟡 @AnthropicAI: The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the Uni — score 50
Sources: twitter_rss
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we
🟡 @GoogleDeepMind: Our Robotics Accelerator has launched with 15 startups helping shape the future of physical AI in Europe. 🤖 This three-month program will connect them with access to our AI stack, Gemini Robotics mod — score 50
Sources: twitter_rss
Our Robotics Accelerator has launched with 15 startups helping shape the future of physical AI in Europe. 🤖 This three-month program will connect them with access to our AI stack, Gemini Robotics models and hands-on support from our teams. Meet the companies → https://goo.gle/4oeEk2K
🟡 @simonw: I got fed up of waiting for OpenAI to bring their much improved gpt-realtime-2 voice conversation model to the ChatGPT product, so I upgraded my OpenAI-WebRTC playground tool to use it and to let you — score 50
Sources: twitter_rss
I got fed up of waiting for OpenAI to bring their much improved gpt-realtime-2 voice conversation model to the ChatGPT product, so I upgraded my OpenAI-WebRTC playground tool to use it and to let you paste in a document to have a conversation about, too https://simonwillison.net/2026/Jun/12/openai-w
Developer Tools
🟡 I am at a hackathon and building a Strategic CMO-cofounder agent. Anyone who wants to try it nowish? — score 67
Sources: reddit/r/AIAgents
I can DM you the link. Would be great to get feedback and questions before judges (in next 60 mins)
🟡 We put 7 LLM agents in a FIFA World Cup betting arena. They are forced to pick a side. (Here is how it works) — score 67
Sources: reddit/r/AIAgents
We're running 7 models against Polymarket's World Cup markets (paper capital, real prices) and some design decisions might interest people building agent evals. The core problem: LLMs are trained to hedge. Ask one "who wins France vs Brazil" and you get a balanced essay. So the protocol forces a dec
🟡 PaddleOCR (v3/v4/v5/v6) implemented in C++ with ncnn [P] — score 56
Sources: reddit/r/MachineLearning
Hi, About a year ago I shared my PaddleOCR implementation here. Since then I've made many improvements, and it now supports PP-OCR v3 through the latest v6 models. The official Paddle C++ runtime has a lot of dependencies and is very complex to deploy. To keep things simple I use ncnn for infere
🟡 LMCache/LMCache — LMCache: Supercharge Your LLM with the Fastest KV Cache Layer — score 46
Sources: github_trending
LMCache: Supercharge Your LLM with the Fastest KV Cache Layer
🟡 We’re building Leangetic ! A local-first compiler for making AI agents cheaper without changing their behavior — score 44
Sources: reddit/r/AIAgents
Hey everyone, We’ve been working on Leangetic, a tool for teams building AI agents that are starting to feel expensive, slow, or hard to control in production. The basic idea is simple: Most agents use an LLM for everything, even when part of the workflow is really just deterministic software wo
Omitted 1 additional developer tools items from the main section; see raw data and source-specific sections below.
Infrastructure & Compute
🟡 Local LLMs aren't democratic anymore... the hardware barrier has gotten out of hand. — score 68
Sources: reddit/r/LocalLLaMA
When we first started experimenting with local LLMs, it was a completely different story! We were using gaming GPUs to tinker around. 8GB or 16GB of VRAM (which wasn't even a given for everyone) was the norm, and so many people could actually get their hands dirty and experiment. Let’s just forget f
Research Papers
🟡 See What I See, Know What I Think: Dense Latent Communication Across Heterogeneous Agents — score 55
Sources: huggingface
Multi-agent systems communicate mostly through text, paying a lossy and expensive decode and re-encode cost. KV-cache communication is a promising alternative, yet most prior work is homogeneous, using duplicate copies of the same model, and avoids the central challenge of cross-model latent alignme
Other Signals
🟡 Statement on the US government directive to suspend access to Fable 5 and Mythos 5 — score 61
Sources: reddit/r/LocalLLaMA
🟡 Slightly reducing the sloppiness of AI generated front end — score 58
Sources: hackernews
🟡 I scaled test-time compute for Qwen-3.6-27B and Gemma-4-31B to surpass Claude Mythos in code optimizations and speedups. — score 46
Sources: reddit/r/LocalLLaMA
The scaffold uses ~25-40x more compute on the original baseline model to attempt the same problem. I put it into max mode by setting the branches exploration breadth to 5, iterative corrections loop depth to 10 and 6 branch aware selective hypothesis that are revised after every 2 iterations. These
🟡 Shepherd's Dog: A Game by the Most Dangerous AI Model — score 42
Sources: hackernews
🟢 Incremental
Model Releases
🟢 New model on huggingface — score 39
Sources: reddit/r/LocalLLaMA
https://huggingface.co/prefeitura-rio/Rio-3.5-Open-397B A qwen finetune. Looks pretty even with qwen 3.7 plus, except it's actually open source. Disclosure: I work as a researcher for the city government of Rio de Janeiro, which developed th
🟢 when fable gets banned but it's ok because you've about to download qwen3.7_67b_21a_mythos_father_fable_mother_distilled_ablated_ablitereted_uncensored_agi_sparse_attention_MTP_SuperHOT_q6_maybe_q7_AGI_FINAL.gguf from huggingface — score 32
Sources: reddit/r/LocalLLaMA
title
🟢 Launch HN: BitBoard (YC P25) – Analytics Workspace for Agents — score 25
Sources: hackernews
🟢 ZONOS2: real-time TTS with 8B params, 900M active, and high-fidelity voice cloning — score 4
Sources: reddit/r/LocalLLaMA
https://reddit.com/link/1u4lk5c/video/kyhdw0uog07h1/player Links: * Blog: https://zyphra.com/our-work/zonos2 * Weights: https://huggingface.co/Zyphra/ZONOS2 * Inference code: [https://github.com/Zyphra/ZONOS2](https://gith
Developer Tools
🟢 basicmachines-co/basic-memory — AI conversations that actually remember. Never re-explain your project to your AI again. Join our Discord:https://discord.gg/tyvKNccgqN — score 35
Sources: github_trending
AI conversations that actually remember. Never re-explain your project to your AI again. Join our Discord:https://discord.gg/tyvKNccgqN
🟢 cocoindex-io/cocoindex — Incremental engine for long horizon agents 🌟 Star if you like it! — score 27
Sources: github_trending
Incremental engine for long horizon agents 🌟 Star if you like it!
🟢 I put my AI agent governance platform online. Try to break it. — score 24
Sources: reddit/r/AIAgents
I’ve spent the last several months building Bendex Arc, a governance layer that sits between AI agents and the real world. As agents get browser access, tools, MCP servers, memory, and the ability to take actions, I kept running into the same gap: nothing was tracking what authority those agents sho
🟢 modelscope/ms-swift — Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.6, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM4.5v, Gemma4, Llava, Phi4, ...) (AAAI 2025). — score 23
Sources: github_trending
Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.6, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM4.5v, Gemma4, Llava, Phi4, ...) (AAAI 2025).
🟢 I'm hiring someone obsessed with AI and the creator economy to scale our membership agency — score 22
Sources: reddit/r/AIAgents
Hi everyone, I'm looking for a strong operator who is equally obsessed and ahead of the curve when it comes to production-ready AI agents for creator services. After using AI heavily since 2022 (and yet, not as much as I'd like), I have a strong sense of the architecture and want to collab with some
Omitted 5 additional developer tools items from the main section; see raw data and source-specific sections below.
Infrastructure & Compute
🟢 colinhacks/zod — TypeScript-first schema validation with static type inference — score 25
Sources: github_trending
TypeScript-first schema validation with static type inference
🟢 Derivative-Free Neural Network Optimization: MNIST Case [R] — score 12
Sources: reddit/r/MachineLearning
A direct optimization test was conducted on a neural network for MNIST image classification. The network features a 784-32-10 architecture with a total of 25,450 continuous parameters (weights and biases). Instead of employing backpropagation or gradient information, the parameters were optimized us
Research Papers
🟢 A Stationary (and Therefore Compatible) Representation is All You Need — score 20
Sources: huggingface
Learning compatible representations aims to learn feature representations that can be used interchangeably over time whenever a model undergoes updates. In this paper, we demonstrate that stationary representations learned by d-Simplex fixed classifiers imply compatibility as in its formal definitio
Other Signals
🟢 GLM-5.2 next week, open weight, MIT — score 25
Sources: reddit/r/LocalLLaMA
🟢 A friendly reminder that APIs are rented, local weights are forever — score 18
Sources: reddit/r/LocalLLaMA
Anthropic just nuked Fable 5 globally to comply with a sudden US export ban because they couldn't instantly verify the nationality of their cloud users. Let this be a massive reminder to the community: if the weights aren't running on your own hardware, you don't actually control your AI. Cloud APIs
🟢 Fable 5 data, including CoT — score 11
Sources: reddit/r/LocalLLaMA
https://huggingface.co/datasets/Glint-Research/Fable-5-traces A simple dataset of all the Fable 5 data we could get our hands on before it was taken away (no clue if it's coming back). Expect some fine-tuned models trained on this soon
📈 Trending Repos
| Repo | Description | Stars Today | Language |
|---|---|---|---|
| shuvonsec/claude-bug-bounty | AI-powered bug bounty hunting from your terminal - recon, 20 vuln classes, autonomous hunting, and report generation. All inside Claude Code. | 226 | python |
| browser-use/browser-use | 🌐 Make websites accessible for AI agents. Automate tasks online with ease. | 181 | python |
| LMCache/LMCache | LMCache: Supercharge Your LLM with the Fastest KV Cache Layer | 28 | python |
| elie222/inbox-zero | The world's best AI personal assistant for email. Open source app to help you reach inbox zero fast. | 27 | typescript |
| basicmachines-co/basic-memory | AI conversations that actually remember. Never re-explain your project to your AI again. Join our Discord:https://discord.gg/tyvKNccgqN | 20 | python |
| cocoindex-io/cocoindex | Incremental engine for long horizon agents 🌟 Star if you like it! | 15 | rust |
| colinhacks/zod | TypeScript-first schema validation with static type inference | 14 | typescript |
| modelscope/ms-swift | Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.6, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM4.5v, Gemma4, Llava, Phi4, ...) (AAAI 2025). | 11 | python |
| novuhq/novu | The open-source communication infrastructure for agents and products | 4 | typescript |
📄 New Papers
| Title | Category | Hotness | Link |
|---|---|---|---|
| See What I See, Know What I Think: Dense Latent Communication Across Heterogeneous Agents | research_paper | 4 | Open |
| A Stationary (and Therefore Compatible) Representation is All You Need | research_paper | 3 | Open |
🐦 Twitter/X Highlights
| Account | Tweet Summary |
|---|---|
| AnthropicAI | The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we Post |
| GoogleDeepMind | Our Robotics Accelerator has launched with 15 startups helping shape the future of physical AI in Europe. 🤖 This three-month program will connect them with access to our AI stack, Gemini Robotics models and hands-on support from our teams. Meet the companies → https://goo.gle/4oeEk2K Post |
| simonw | I got fed up of waiting for OpenAI to bring their much improved gpt-realtime-2 voice conversation model to the ChatGPT product, so I upgraded my OpenAI-WebRTC playground tool to use it and to let you paste in a document to have a conversation about, too https://simonwillison.net/2026/Jun/12/openai-w Post |
Repeated From Recent Briefings
- farion1231/cc-switch — A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Gemini CLI & Hermes Agent. Only official website: ccswitch.io - first seen 2026-05-08
- NVIDIA/SkillSpector — Security scanner for AI agent skills. Detect vulnerabilities, malicious patterns, and security risks. - first seen 2026-06-10
- Risk Under Pressure: Compute-Aware Evaluation of Adversarial Robustness in Language Models - first seen 2026-06-11
- Spent $3 running 4x4090 benchmarks for llama 3 70b (exl2 vs gguf). exl2 generation speed is kind of ridiculous. - first seen 2026-06-12
- anomalyco/opencode — The open source coding agent. - first seen 2026-05-09
- maziyarpanahi/openmed — open-source healthcare ai - first seen 2026-06-10
- anthropics/skills — Public repository for Agent Skills - first seen 2026-05-11
- Rethinking Psychometric Evaluation of LLMs: When and Why Self-Reports Predict Behavior - first seen 2026-06-12
- FareedKhan-dev/train-llm-from-scratch — A straightforward method for training your LLM, from downloading data to generating text. - first seen 2026-06-11
- Are AI agents making traditional software interfaces obsolete? - first seen 2026-06-12
- ... plus 34 more repeated items in processed data