Max Headroom

@droidbuilds diff is 8

most people get this wrong what’s the difference between 100 MB/s and 100 Mb/s?

@LenSeaside @LottoLabs I am. You don't understand how tool calling works, are you sure you want to keep adding to this discussion?

@CosmicMonad @LottoLabs I assumed you would be trying to get the best out of it.

This seems like it might be the best local model of the Gemma family Maybe better than qwen 27b? huggingface.co/Jiunsong/Super…

@x0ptimal @bnjmn_marie I haven't really used the MoE qwen3.5 offerings, because the dense works so well. Glad they are also functional!

@CosmicMonad @bnjmn_marie Can agree from experience. Its straight dog water when it comes to tool calling. Been running qwen3.5 35b and it just works perfectly 👌

I published my full analysis comparing Gemma 4 31B vs Qwen3.5 27B. > Best accuracy: Gemma 4 31B > Best token efficiency: Gemma 4 31B > Best raw inference throughput: Qwen3.5 27B > Best memory footprint: Qwen3.5 27B > Best end-to-end latency: task-dependent, with Gemma 4 slightly ahead on harder tasks and Qwen3.5 ahead on simpler ones > Best “fast” mode: Gemma 4 31B with thinking disabled > Best generalization / least benchmark affinity: Gemma 4 31B All the results and data here: kaitchup.substack.com/p/gemma-4-31b-…

its crazy how nobody talks about the epstein files anymore

@venusanbacchus Why does this graph only go to 130?

i got an iq test done and then i saw a graph like this a few months later and stopped caring

@Logically_JC Every 4 years literally half the country stands against the current president, and gets its way roughly 50% of the time. Not that it matters to the epstein class, they run both candidates who answer to them.

Yes, you can.

@EvanAKilgore What, just now he does? Seriously?

HOLY SH*T: Marjorie Taylor Greene just revealed her kids received serious death threats after Trump called her a "traitor." She reached out to Trump directly. President Trump criticized her and said it's het fault if her son gets killed. Bro... Donald Trump sounds evil now.

Dense please sirs?

@__alula I hate gnome fuckers, what a festering cloaca of useless shit.

GNOME FOR FUCKS SAKE WHY

@InnoboSJ @Teknium @TheRealMarcosV Yea it does :)

@Teknium @TheRealMarcosV Qwen 27B doesn't support vision I don't believe, but the 4B and 9B variants do.

Qwen 27B seems to be Twitter's choice for better local model than Gemma-4

@Teknium @TheRealMarcosV Wat? qwen3.5-27b *IS* a vision model... And it works in hermes just fine.

@TheRealMarcosV You need to set a vision model (I dont think qwen has vision)

@Alibaba_Qwen Awesome!! Thank you Qwen team! Hope very much Qwen3.6-27B is also coming soon!! You guys are awesome!

⚡ Meet Qwen3.6-35B-A3B：Now Open-Source！🚀🚀 A sparse MoE model, 35B total params, 3B active. Apache 2.0 license. 🔥 Agentic coding on par with models 10x its active size 📷 Strong multimodal perception and reasoning ability 🧠 Multimodal thinking + non-thinking modes Efficient. Powerful. Versatile. Try it now👇 Blog：qwen.ai/blog?id=qwen3.… Qwen Studio：chat.qwen.ai HuggingFace：huggingface.co/Qwen/Qwen3.6-3… ModelScope：modelscope.cn/models/Qwen/Qw… API（‘Qwen3.6-Flash’ on Model Studio）：Coming soon～ Stay tuned

@AburageFox @bnjmn_marie Языковые показатели у гэмы действительно выше, и если тебе важен обычный чат бот, то вполне можно разговаривать с гэмой по японски. Для реальной работы гэма бесполезна.

@CosmicMonad @bnjmn_marie Qwen3.5 27Bは日本語能力がgemma4 26BA4Bと比較しても明確に劣るから、日本人視点だとgemma4の評価は高いですよ Qwen3.5の日本語は常に微妙に話がズレるし、言語の解釈や認識もガバガバなんだ……

@theo Why don't people just use Cline? How can you even get some bot to properly deal with a code base? By the time you finish writing all the possible guidelines to make it somewhat not lose its brain and spaghettify your code, you might as well have done it yourself.

Was I right?

@Cointelegraph The word professor should have been in quotes too.

🇨🇳 TODAY: Chinese professor Jiang Xueqin says Bitcoin is a "CIA operation," questioning where the blockchain servers are located.

@LottoLabs Is that listing prefill anywhere? And wow, 14t/s would be painfully slow. I mean, for playing around could be fun, for work... not so much.

This is so ass, 14t/s Stick to your 3090

@im_yeyito @bnjmn_marie Gemma4 fails at tool calls, constantly. It is not a viable model for work. Qwen3.5-27b is. This is an ad campaign to try and make it seem that local LLM field is empty, once this trash gemma4 is made in public eyes seem like it is the top model, so no one even tries using 27b.

@bnjmn_marie this is exactly the kind of comparison people need more of. not "which model wins one screenshot benchmark" but the actual tradeoff surface: accuracy, throughput, token efficiency, memory footprint, end-to-end latency. that is how you pick a model for real work

@MemoryReboot_ @bnjmn_marie Bullshit.

@bnjmn_marie Great job Looks like 27b dethroned finally

@mudscryer You haven't figured out yet that you shouldn't be public about this? Hmm.

It’s really awful being 143 IQ and such a total failure in life. I got my PhD in biology thinking I was going to help save the world but no one has hired me for a permanent position in over a year and a half of searching. I thought proof of my high IQ was supposed to land me a job