NullFoundry @nullfoundry

btw, I find what AI Labs are doing disgusting. I won't name names, but they're digitizing books and destroying them all. disgusting

@sachindetrax @bijanbowen sad man

Opinion / Thoughts: I wonder when we'll get a model better than Qwen 3.6 27B. I have seen a lot of benchmarks (even though I think every lab is benchmaxxing), but I really think the real test is doing some code, testing tooling, real stress tests. @bijanbowen is doing an amazing job recording videos with real tests. And I want see more people doing the same. While I think comparing speed is an important thing, it doesn't matter how fast it is if a model is "dumb." GPT Sol and Opus 4.8-5/Fable are doing a great job. Even though they are closed models, I'm currently using them myself in my projects ( personal and current job ), interchanging them between Qwen3.6. I am using Qwen 3.6 27B Q4_KM but recently switched to Q4_NL because it seems better and faster. However, I've seen people getting better results with Q5+ quantizations. Unfortunately, even though I bought 2x 3090s, I haven't had the time to build this new rig to add up to 48GB RAM. It's not just a matter of time, but also the lack of availability and the overpriced parts here in Brazil. Now, running local models it's not for everyone unfortunately. Ram and GPU are incredible expensive. We don't know for sure if the issue is manufactures controlling prices, meaning corruption, or in fact, these guys are not able to supply the demand. I see guys purchasing DX sparks, RTX 6000, and don't think this is affordable in fact. Consumers are not be able to buy freely these parts. A DX Spark costs $4,699 in America, but in Brazil, it’s R$ 64,999.99, which is almost 3 times more expensive ( even though the conversion rate is 1$ <> 5.12). This happens because the government charges a lot of taxes, and sellers here have a culture of making a big profit instead of profiting from volume. So, at the end of the day, I hope we have better quantizations or an equation that solves the issue of having massive models. To me, it would be great if in the future we could have models like qwen3.6 27b with the intelligence of a fable (idk if this is possible). Either way, I'm still excited about local models, and I will keep pursuing finishing my 2x 3090 rig and studying this world. So, what do you think guys? Am I wrong? I want to hear your thoughts.

Qwen3.8 is launching and going open-weight soon!🌐 With a massive 2.4T parameters, this model is continuously evolving. We believe it’s one of the most powerful model available today, compatible to leading frontier AI models , second only to Fable 5. You don't have to wait to test it. Just now, the Qwen3.8-Max-Preview made its debut on Alibaba’s Token Plan, Qoder, and QoderWork. Be among the very first to try it out. Can't wait to hear what you build. Stay tuned! 🚀  Token Plan international：qwencloud.com/pricing/token-… China：platform.qianwenai.com/pricing/token-…

Kimi K3 subway FPS result, some say it's benchmaxxed due to my influencerstatus™️ idk man, idk...

nvm my auto updater is broken lmao

hey guys, I'm a bit off local models these days because I was testing ChatGPT with MCP, and I updated my llama.cpp to version b10000, but I'm only getting 300 t/s on prefill. I haven't noticed any changes on my end. Is anyone else experiencing the same?

@OpenAI chatgpt is doooooown

@Hikari_07_jp Yeah its annoying

Tip of the day You can expose your Hermes agent over the MCP protocol and connect it to ChatGPT. This repo makes this possible: github.com/asimons81/herm… But be careful because this can lead to some seriously bad consequences. In my case, I made a small OAuth server proxy, so I'm protecting everything on my side. I am using ngrok as a tunnel. Downside: you don't have the history within hermes agent Upside: you don't spend tokens and chatgpt has more context of your stuff. Have you tried this before?

ngl, lots of AI posts are leading to misunderstandings in this world of local models. Some guys don't even check the facts before posting them

@HuggingModels bruh

Gonna need to test it

Hangoo Kang @hangoo_kang

4 weeks ago

“TRACE: Capability-Targeted Agentic Training” got Spotlight @ ICML AIWILD 🎉 Beats direct RL, GEPA, & synthetic-agent data on SWE-Bench Verified and τ²-Bench. TRACE-Qwen3.6-27B tops GPT-5.2-Codex, GLM 5, & Claude 4.5 Sonnet on SWE-Bench. Co-led with @TarunSures41845. Thanks to

14 36 235 170K 241

“TRACE: Capability-Targeted Agentic Training” got Spotlight @ ICML AIWILD 🎉 Beats direct RL, GEPA, & synthetic-agent data on SWE-Bench Verified and τ²-Bench. TRACE-Qwen3.6-27B tops GPT-5.2-Codex, GLM 5, & Claude 4.5 Sonnet on SWE-Bench. Co-led with @TarunSures41845. Thanks to @JonSaadFalcon and our advisor @Azaliamirh. Details below 👇

@Hermesage make gpt live possible and you will gain my loyalty

I am very impressed with Grok 4.5. I'm getting good results, pretty consistent. This video from @bijanbowen shows some impressive things. The car game is pretty good, btw. youtu.be/DYDF_UZc0n4

🖥️Damn Local AI community. Google comes through. Gemma 4 is now fully on-device in React Native! 🎯 Run Google’s Gemma 4 100% offline in your cross-platform mobile apps with hardware acceleration: ⚡ Vulkan delegate on Android ⚡ MLX delegate on Apple Silicon 🏃‍♂️How to get started: ✦︎ Use the open-source library: react-native-executorch ✦︎ Load models easily with the useLLM hook ✦︎ Supports vision + tool calling locally 💰Cost: ✦︎ $0 API cost — fully on-device ✦︎ Main cost = device performance (best on newer phones) Best for: ✦︎ Vibe-coded React Native apps that need strong local AI with privacy and no internet dependency. 🔗 Demo & Docs: github.com/software-mansi…

Google Gemma @googlegemma

3 weeks ago

Gemma 4 now works on-device using React Native! You can now run Gemma 4 fully offline in your cross-platform apps with local hardware acceleration: ⚡ Vulkan delegate on Android ⚡ MLX delegate on Apple Silicon See Gemma 4's vision and tool-use capabilities in action, instantly

33 173 1K 116K 839

@akshay_pachaar Is it possible use it on llama cpp?

@TheExpanseRPG just let me play it