And here we go again. A new stealth model.

There's a new stealth model in Cursor from one of our partners. It's free to try, let us know what you think!

New stealth model in Cline: "Sonic" Designed for coding & free to use, because your usage helps improve the model while it's in alpha.

reddit.com/r/StableDiffus…

where is the model card deepseek lol

Hey Logan Kilpatrick Josh Woodward It would be cool if we could use Gemini Live (at least voice real time chat) with Gemini Web, right? We can already do this with the competition...

its a very boring google live lol

where is my nano banana model

wow mini high is almost same bench score as gpt-5 medium

Introducing #ZenFlow: No Compromising Speed for #LLM Training w/ Offloading 5× faster LLM training with offloading 85% less GPU stalls 2× lower I/O overhead 🚀 Blog: hubs.la/Q03DJ6GJ0 🚀 Try ZenFlow and experience 5× faster training with offloading: hubs.la/Q03DJ6Vb0

ok now kimi turns to be much more interesting

Introducing DeepSeek-V3.1: our first step toward the agent era! 🚀 🧠 Hybrid inference: Think & Non-Think — one model, two modes ⚡️ Faster thinking: DeepSeek-V3.1-Think reaches answers in less time vs. DeepSeek-R1-0528 🛠️ Stronger agent skills: Post-training boosts tool use and

🚀 In March, we launched Paper Finder, an LLM-powered literature search agent that surfaces papers other tools miss. Now, we’re releasing an open-source snapshot to enable others to inspect & build on it—and reproduce the results. 🧵

Meet PACT, my new head-to-head negotiation benchmark for LLMs! Each game runs 20 rounds; in every round the agents exchange a brief chat, then post a bid and an ask. A trade clears at the midpoint when bid ≥ ask. From 5,000+ matchups across 30 models, GPT-5 leads. 🏆

Introducing Mirage 2 — a real-time, general-domain generative world engine you can play online Upload any image—photos, concept art, classic paintings, kids' drawings—and step into it as a live, interactive world. Prompt your worlds with text to create any surreal scenes and

It's better to mess with a codebase of thousands of files than to mess with Comfyui nodes.

With over 200 million tokens over the last few days put into Qwen 3 Coder provider testing. I finally have results I can share. 1. I think we need to be able to multi-select providers in any ai coding tool. 2. API reliability was a lot bigger of a problem with Qwen 3 Coder than

Veo 3 for free this weekend. Three videos for free.

Now all that's left is for Qwen to launch a TTS/ASR multilingual SOTA, and then they will have mastered all AI OSS.

Finally!