Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Numbers in! Target: 150 tok/s browser-native Qwen 3.5 inference Achieved: 180 tok/s — Qwen 3.5 INT4, WebGPU, browser-only. No cloud, APIs, servers, just Pure WebGPU + WGSL optimizations, HF SafeTensors, single browser tab. Open AI, for real.

Ex0byt's profile picture

Eric

11,672 subscribers

256,656 Aufrufe • vor 6 Monaten •via X (Twitter)

BildungWissenschaft & Technologie
LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

Qwen-2.5 on WebGPU 🏎️ • 42 tok/sec for Qwen2.5-Coder-1.5B on Mac ⚡ • Powered by MLC WebLLM and WebGPU 🔥 Watch Qwen2.5-Coder-1.5B build a website entirely in the browser!
0:35

Qwen-2.5 on WebGPU 🏎️ • 42 tok/sec for Qwen2.5-Coder-1.5B on Mac ⚡ • Powered by MLC WebLLM and WebGPU 🔥 Watch Qwen2.5-Coder-1.5B build a website entirely in the browser!

Caleb

41,579 Aufrufe • vor 1 Jahr

NEW: Alibaba just released Qwen 3.5 Small — a family of powerful multimodal models available in a range of sizes (0.8B, 2B, 4B, and 9B parameters). Perfect for on-device applications! They can even run 100% locally in your browser on WebGPU, powered by Transformers.js! 🤯
0:46

NEW: Alibaba just released Qwen 3.5 Small — a family of powerful multimodal models available in a range of sizes (0.8B, 2B, 4B, and 9B parameters). Perfect for on-device applications! They can even run 100% locally in your browser on WebGPU, powered by Transformers.js! 🤯

Xenova

102,592 Aufrufe • vor 6 Monaten

This doesn't feel like it should be possible. Muse Glimmer 30B can now run locally in your browser on WebGPU... at the same speed as llama.cpp (~25 tok/s on my M4 Max)! Just open a website. No installation required.
1:22

This doesn't feel like it should be possible. Muse Glimmer 30B can now run locally in your browser on WebGPU... at the same speed as llama.cpp (~25 tok/s on my M4 Max)! Just open a website. No installation required.

Xenova

21,313 Aufrufe • vor 25 Tagen

Introducing Phi-3 WebGPU, a private and powerful AI chatbot that runs locally in your browser, powered by 🤗 Transformers.js and onnxruntime-web! 🔒 On-device inference: no data sent to a server ⚡️ WebGPU-accelerated (> 20 t/s) 📥 Model downloaded once and cached Try it out! 👇
1:26

Introducing Phi-3 WebGPU, a private and powerful AI chatbot that runs locally in your browser, powered by 🤗 Transformers.js and onnxruntime-web! 🔒 On-device inference: no data sent to a server ⚡️ WebGPU-accelerated (> 20 t/s) 📥 Model downloaded once and cached Try it out! 👇

Xenova

117,724 Aufrufe • vor 2 Jahren

A full x86_64 Linux VM sandbox in WASM with host OS read-only folder mount, accessible via native tool calls by a WebGPU LLM. Video is real time. There is no server, no privacy concerns, no “escapes”. Comfortably in a browser tab. Sandboxes work. If they don’t: Not a sandbox.
1:34

A full x86_64 Linux VM sandbox in WASM with host OS read-only folder mount, accessible via native tool calls by a WebGPU LLM. Video is real time. There is no server, no privacy concerns, no “escapes”. Comfortably in a browser tab. Sandboxes work. If they don’t: Not a sandbox.

remy🐀

43,729 Aufrufe • vor 26 Tagen

BOOM! 💥 Today I added WebGPU support for Andrej Karpathy's nanochat models, meaning they can run 100% locally in your browser (no server)! The d32 version runs at over 50 tps on my M4 Max 🚀 Pretty wild that you can now deploy AI applications using just a single index.html file 😅
0:20

BOOM! 💥 Today I added WebGPU support for Andrej Karpathy's nanochat models, meaning they can run 100% locally in your browser (no server)! The d32 version runs at over 50 tps on my M4 Max 🚀 Pretty wild that you can now deploy AI applications using just a single index.html file 😅

Xenova

95,481 Aufrufe • vor 10 Monaten

🚨 Open-source AI just got serious. Alibaba just released Qwen 3.5-Plus 🤯 397B parameters. Only 17B active per inference. Performance comparable to Gemini 3 Pro — at just $0.8 per million tokens (1/18 the cost). But here’s what makes it different 👇 This isn’t just another benchmark drop. In the demo, Qwen runs directly inside the terminal: ➙ Reviews GitHub PRs ➙ Fetches diffs automatically ➙ Runs CLI & smoke tests ➙ Generates integration tests ➙ Executes /context ➙ Shows real-time token usage breakdown That’s not a chatbot. That’s a real developer workflow. Under the hood: ⚡ 60% lower deployment memory ⚡ Inference throughput increased by 8.6× (32K context) ⚡ Inference throughput increased by up to 19× (256K ultra-long context) ⚡ Native multimodal model ⚡ Supports direct video input (up to 2 hours) And the ecosystem is massive: ✔ 201 languages ✔ 250K vocabulary ✔ 400+ open-sourced models ✔ 1B+ downloads ✔ 200K+ derivative models Available now via Qwen App, MoDa Community, Hugging Face, and Alibaba Cloud Bailian. ✅ The AI race isn’t about who’s biggest anymore. → It’s about who’s usable. 🔥 Watch the demo 👇 #Qwen #OpenSourceAI #AI #DevTools
1:10

🚨 Open-source AI just got serious. Alibaba just released Qwen 3.5-Plus 🤯 397B parameters. Only 17B active per inference. Performance comparable to Gemini 3 Pro — at just $0.8 per million tokens (1/18 the cost). But here’s what makes it different 👇 This isn’t just another benchmark drop. In the demo, Qwen runs directly inside the terminal: ➙ Reviews GitHub PRs ➙ Fetches diffs automatically ➙ Runs CLI & smoke tests ➙ Generates integration tests ➙ Executes /context ➙ Shows real-time token usage breakdown That’s not a chatbot. That’s a real developer workflow. Under the hood: ⚡ 60% lower deployment memory ⚡ Inference throughput increased by 8.6× (32K context) ⚡ Inference throughput increased by up to 19× (256K ultra-long context) ⚡ Native multimodal model ⚡ Supports direct video input (up to 2 hours) And the ecosystem is massive: ✔ 201 languages ✔ 250K vocabulary ✔ 400+ open-sourced models ✔ 1B+ downloads ✔ 200K+ derivative models Available now via Qwen App, MoDa Community, Hugging Face, and Alibaba Cloud Bailian. ✅ The AI race isn’t about who’s biggest anymore. → It’s about who’s usable. 🔥 Watch the demo 👇 #Qwen #OpenSourceAI #AI #DevTools

Toha Khan

123,375 Aufrufe • vor 6 Monaten

🤯 A localmaxxer hit ~381 tok/s on a SINGLE RTX 3090 with Qwen3.8-27B. This developer has turned a 24GB RTX 3090 into a monster Qwen inference box - w/some creativity. Four days ago 👉 ⚡ ~82 tok/s single-user Then 👉 ⚡ ~114 tok/s with optimized MTP ⚡ ~138 tok/s with DFlash2 + lookup drafting Now 👉 🔥 ~381 tok/s on ONE request How? The recipe combines ... 🧠 Qwen3.8-27B 🎮 1× RTX 3090 24GB @ 250W ⚙️ heavily optimized vLLM ⚡ DFlash2 speculative decoding 🔎 lookup-augmented drafting 📚 prefix caching 🧮 16-token verification blocks 💾 quantized KV / heads / activations DFlash2 normally proposes 7 tokens. The developer realized the verification block doesn't have to stop there. If Qwen is answering from a document already sitting in the prompt, the system can fill the remaining draft positions using tokens found directly in that context. 🎯 So the target model can verify 16 tokens at once. On a ~25K-token document reproduction task: Previous DFlash2 👉 ~260 tok/s Longer verification + context lookup 👉 🔥 ~382 tok/s Acceptance: 🤯 15 of 16 tokens per verification step That is where the crazy number comes from. ⚠️ On ordinary real-world chat prompts, the same setup is around ~133 tok/s Still extremely fast for a dense 27B model on an RTX 3090. The 381 tok/s mode shines when the answer largely comes from material already in context so these are best use cases 📚 RAG / document Q&A 💻 Coding assistants applying edits 📝 Quoting or rewriting documents 🔎 Extracting information from long prompts And another optimization 👉 With prefix caching, a second question against the same 25K-token document reportedly goes from: 🐌 22.4 sec TTFT → ⚡ 0.56 sec TTFT Because the model doesn't need to process the whole document again. 🎯 It's specifically a mode for RAG front ends and coding agents. Follow iamMess on Reddit or syv-ai on GitHub 🔗 Reddit: r/LocalLLaMA/comments/1vtup5s/ 🔗 GitHub: /syv-ai/qwen38-27b-rtx3090
1:08

🤯 A localmaxxer hit ~381 tok/s on a SINGLE RTX 3090 with Qwen3.8-27B. This developer has turned a 24GB RTX 3090 into a monster Qwen inference box - w/some creativity. Four days ago 👉 ⚡ ~82 tok/s single-user Then 👉 ⚡ ~114 tok/s with optimized MTP ⚡ ~138 tok/s with DFlash2 + lookup drafting Now 👉 🔥 ~381 tok/s on ONE request How? The recipe combines ... 🧠 Qwen3.8-27B 🎮 1× RTX 3090 24GB @ 250W ⚙️ heavily optimized vLLM ⚡ DFlash2 speculative decoding 🔎 lookup-augmented drafting 📚 prefix caching 🧮 16-token verification blocks 💾 quantized KV / heads / activations DFlash2 normally proposes 7 tokens. The developer realized the verification block doesn't have to stop there. If Qwen is answering from a document already sitting in the prompt, the system can fill the remaining draft positions using tokens found directly in that context. 🎯 So the target model can verify 16 tokens at once. On a ~25K-token document reproduction task: Previous DFlash2 👉 ~260 tok/s Longer verification + context lookup 👉 🔥 ~382 tok/s Acceptance: 🤯 15 of 16 tokens per verification step That is where the crazy number comes from. ⚠️ On ordinary real-world chat prompts, the same setup is around ~133 tok/s Still extremely fast for a dense 27B model on an RTX 3090. The 381 tok/s mode shines when the answer largely comes from material already in context so these are best use cases 📚 RAG / document Q&A 💻 Coding assistants applying edits 📝 Quoting or rewriting documents 🔎 Extracting information from long prompts And another optimization 👉 With prefix caching, a second question against the same 25K-token document reportedly goes from: 🐌 22.4 sec TTFT → ⚡ 0.56 sec TTFT Because the model doesn't need to process the whole document again. 🎯 It's specifically a mode for RAG front ends and coding agents. Follow iamMess on Reddit or syv-ai on GitHub 🔗 Reddit: r/LocalLLaMA/comments/1vtup5s/ 🔗 GitHub: /syv-ai/qwen38-27b-rtx3090

David Hendrickson

111,228 Aufrufe • vor 16 Tagen

i spent 3 hours finding the sweet spot for hermes 4.3 36B on a single RTX 3090. saving you the trouble, anon. the model is 21.8GB at Q4_K_M. that leaves 2.2GB free on 24GB VRAM. not much room for KV cache. here's what actually happened: 4K: 35.3 tok/s 8K: 35.2 tok/s 16K: 34.8 tok/s 32K: 34.6 tok/s 64K: 6.4 tok/s 128K: 1.9 tok/s flat from 4K to 32K. then it falls off a cliff at 64K. the trick is quantized KV cache. without it you OOM at 16K. with quantized KV cache you get 32K at full speed. all 64 layers on GPU. at 64K something weird happens. ngl 99 (all layers on GPU) = 3.96 tok/s. the KV cache silently spills to CPU. drop to ngl 55 and speed jumps to 6.37. drop to ngl 48 and it gets worse again (3.46). there's an offload sweet spot where you free just enough VRAM for the cache without losing too much compute to PCIe transfers. 128K works at ngl 32 but you're at 1.95 tok/s. half the model on CPU. usable for batch work, not for interactive. the sweet spot command: llama-server -m hermes-4.3-36b-Q4_K_M.gguf -ngl 99 -c 32768 --cache-type-k q4_0 --cache-type-v q4_0 32K context. 34.6 tok/s. all on GPU. this is where dense 36B lives on 24GB. for comparison, qwen 3.5 (35B MoE, 3B active) holds 112 tok/s from 4K all the way to 262K on the same GPU. no speed drop. same total params, completely different architecture. hybrid linear attention means flat context scaling. dense pays for every token in the KV cache. code and quality comparison coming next. fast vs slow generation side by side in the videos below.
2:29

i spent 3 hours finding the sweet spot for hermes 4.3 36B on a single RTX 3090. saving you the trouble, anon. the model is 21.8GB at Q4_K_M. that leaves 2.2GB free on 24GB VRAM. not much room for KV cache. here's what actually happened: 4K: 35.3 tok/s 8K: 35.2 tok/s 16K: 34.8 tok/s 32K: 34.6 tok/s 64K: 6.4 tok/s 128K: 1.9 tok/s flat from 4K to 32K. then it falls off a cliff at 64K. the trick is quantized KV cache. without it you OOM at 16K. with quantized KV cache you get 32K at full speed. all 64 layers on GPU. at 64K something weird happens. ngl 99 (all layers on GPU) = 3.96 tok/s. the KV cache silently spills to CPU. drop to ngl 55 and speed jumps to 6.37. drop to ngl 48 and it gets worse again (3.46). there's an offload sweet spot where you free just enough VRAM for the cache without losing too much compute to PCIe transfers. 128K works at ngl 32 but you're at 1.95 tok/s. half the model on CPU. usable for batch work, not for interactive. the sweet spot command: llama-server -m hermes-4.3-36b-Q4_K_M.gguf -ngl 99 -c 32768 --cache-type-k q4_0 --cache-type-v q4_0 32K context. 34.6 tok/s. all on GPU. this is where dense 36B lives on 24GB. for comparison, qwen 3.5 (35B MoE, 3B active) holds 112 tok/s from 4K all the way to 262K on the same GPU. no speed drop. same total params, completely different architecture. hybrid linear attention means flat context scaling. dense pays for every token in the KV cache. code and quality comparison coming next. fast vs slow generation side by side in the videos below.

Sudo su

52,268 Aufrufe • vor 6 Monaten

🚨 do you understand what this AMD engineer just proved on camera watch the clip, it’s all there in one frame: llama.cpp slot logs on the left, nvitop showing 79% GPU util holding steady, localhost chat answering a 26k token prompt at 711 tok/s prompt eval, the gemma-4-12B Q4_K_XL gguf running clean. no cloud call, no api key, no censorship layer. for 2 years the rule was: frontier models need the cloud, local is too slow for real work, pay $200/month and wait for tokens. today a $1,800 lunch box runs qwen 3 235B at 11 tok/s with zero data leaving the room, and the same box runs gemma 4 12B with multi token prediction at 14+ tok/s on 30k context with no throttling. $440/month subscription stack: dead. $200/month claude tier: dead. local is too slow excuse: dead. one lunch box. $9/month in electricity. two moats buried in one week. the you need the cloud line wasn’t protecting users it was protecting the subscription.
0:21

🚨 do you understand what this AMD engineer just proved on camera watch the clip, it’s all there in one frame: llama.cpp slot logs on the left, nvitop showing 79% GPU util holding steady, localhost chat answering a 26k token prompt at 711 tok/s prompt eval, the gemma-4-12B Q4_K_XL gguf running clean. no cloud call, no api key, no censorship layer. for 2 years the rule was: frontier models need the cloud, local is too slow for real work, pay $200/month and wait for tokens. today a $1,800 lunch box runs qwen 3 235B at 11 tok/s with zero data leaving the room, and the same box runs gemma 4 12B with multi token prediction at 14+ tok/s on 30k context with no throttling. $440/month subscription stack: dead. $200/month claude tier: dead. local is too slow excuse: dead. one lunch box. $9/month in electricity. two moats buried in one week. the you need the cloud line wasn’t protecting users it was protecting the subscription.

AdiiX

22,223 Aufrufe • vor 2 Monaten

🚨Architects are going to hate this. Someone just open sourced a full 3D building editor that runs entirely in your browser. No AutoCAD. No Revit. No $5,000/year licenses. It's called Pascal Editor. Built with React Three Fiber and WebGPU -- meaning it renders directly on your GPU at near-native speed. Here's what's inside this thing: → A full building/level/wall/zone hierarchy you can edit in real time → An ECS-style architecture where every object updates through GPU-powered systems → Zustand state management with full undo/redo built in → Next.js frontend so it deploys as a web app, not a desktop install → Dirty node tracking -- only re-renders what changed, not the whole scene Here's the wildest part: You can stack, explode, or solo individual building levels. Select a zone, drag a wall, reshape a slab -- all in 3D, all in the browser. Architecture firms pay $50K+ per seat for BIM software that does this workflow. This is free. 100% Open Source.
1:38

🚨Architects are going to hate this. Someone just open sourced a full 3D building editor that runs entirely in your browser. No AutoCAD. No Revit. No $5,000/year licenses. It's called Pascal Editor. Built with React Three Fiber and WebGPU -- meaning it renders directly on your GPU at near-native speed. Here's what's inside this thing: → A full building/level/wall/zone hierarchy you can edit in real time → An ECS-style architecture where every object updates through GPU-powered systems → Zustand state management with full undo/redo built in → Next.js frontend so it deploys as a web app, not a desktop install → Dirty node tracking -- only re-renders what changed, not the whole scene Here's the wildest part: You can stack, explode, or solo individual building levels. Select a zone, drag a wall, reshape a slab -- all in 3D, all in the browser. Architecture firms pay $50K+ per seat for BIM software that does this workflow. This is free. 100% Open Source.

Guri Singh

2,967,409 Aufrufe • vor 5 Monaten

Someone open-sourced a 66M parameter model that beats ElevenLabs, OpenAI, and Gemini and runs entirely offline on a Raspberry Pi. it's called Supertonic, a text-to-speech model so fast it turns an entire webpage into audio in under 1 second. locally. offline. for free. text-to-speech has lived in the cloud for years. every spoken character was an API call and a fraction of a cent. Supertonic 3 kills that entirely. the model is 99M parameters. ships as a single ONNX file. hits 167× faster than real-time on a laptop CPU. that's ~1,263 characters of speech per second. other open systems sit at 55–287. no cloud. no API. no GPU. → 31 languages, no separate adapters → works inside a browser tab (WebGPU/WASM) → handles phone numbers, currency, dates — no preprocessing → inline tags for laugh and breath → Turns an entire webpage into audio in under 1 second → 44.1kHz studio-grade WAV, no upsampler needed` and on real-world text, "$5.2M", "(212) 555-0142 ext. 402", "30kph”, it's the only one that reads them correctly. ElevenLabs Flash, OpenAI TTS-1, Gemini 2.5 Flash all fail. Supertonic passes. 11.2k stars. 100% Open Source.
1:00

Someone open-sourced a 66M parameter model that beats ElevenLabs, OpenAI, and Gemini and runs entirely offline on a Raspberry Pi. it's called Supertonic, a text-to-speech model so fast it turns an entire webpage into audio in under 1 second. locally. offline. for free. text-to-speech has lived in the cloud for years. every spoken character was an API call and a fraction of a cent. Supertonic 3 kills that entirely. the model is 99M parameters. ships as a single ONNX file. hits 167× faster than real-time on a laptop CPU. that's ~1,263 characters of speech per second. other open systems sit at 55–287. no cloud. no API. no GPU. → 31 languages, no separate adapters → works inside a browser tab (WebGPU/WASM) → handles phone numbers, currency, dates — no preprocessing → inline tags for laugh and breath → Turns an entire webpage into audio in under 1 second → 44.1kHz studio-grade WAV, no upsampler needed` and on real-world text, "$5.2M", "(212) 555-0142 ext. 402", "30kph”, it's the only one that reads them correctly. ElevenLabs Flash, OpenAI TTS-1, Gemini 2.5 Flash all fail. Supertonic passes. 11.2k stars. 100% Open Source.

Superman

30,416 Aufrufe • vor 1 Monat

update: qwen 3.6 27b dense q4 just one shotted octopus invaders game on a single 3090. hermes agent drove the whole thing, ~41 tok/s gen 21gb vram at full 262k context, thinking mode on. one prompt in and the canonical multi-file space shooter benchmark out, the same exact prompt i ran on qwen 3.5 27b dense back in march on the same card. 3.5 needed one external scope bug fix before the game would even load on first play. 3.6 needed nothing. 11 of 11 files written, 2411 lines of code, zero steering interventions, zero external fixes, playable on first load. 16 minutes 41 seconds wall clock from prompt to playable. consumer tier king on a single 3090 is locked tonight, and the silicon underneath my desk did not change between march and now. the open source ecosystem just moved the floor. watch it ship itself, the full 16 minutes 41 seconds sped to 3 minutes 45, no human touched the keyboard between the first prompt and the final frame.
3:45

update: qwen 3.6 27b dense q4 just one shotted octopus invaders game on a single 3090. hermes agent drove the whole thing, ~41 tok/s gen 21gb vram at full 262k context, thinking mode on. one prompt in and the canonical multi-file space shooter benchmark out, the same exact prompt i ran on qwen 3.5 27b dense back in march on the same card. 3.5 needed one external scope bug fix before the game would even load on first play. 3.6 needed nothing. 11 of 11 files written, 2411 lines of code, zero steering interventions, zero external fixes, playable on first load. 16 minutes 41 seconds wall clock from prompt to playable. consumer tier king on a single 3090 is locked tonight, and the silicon underneath my desk did not change between march and now. the open source ecosystem just moved the floor. watch it ship itself, the full 16 minutes 41 seconds sped to 3 minutes 45, no human touched the keyboard between the first prompt and the final frame.

Sudo su

126,487 Aufrufe • vor 3 Monaten

Qwopus on a single RTX 3090. Claude Opus 4.6 reasoning distilled into Qwen 3.5 27B dense, running through Claude's own coding agent (claude code). 29-35 tok/s with thinking mode on. the jinja bug that kills thinking on base Qwen doesn't carry over. harness and model matched. the base model would pause mid task on Claude Code. just stop generating. that's why i ran it through OpenCode, which handles stalled states automatically. this distilled version doesn't stall. it waits for tool outputs, reads them, selfcorrects when something breaks, and keeps going. i gave it a benchmark analysis task. went 9 minutes autonomous. wrote a README nobody asked for. zero steering. video is 5x speed but fully uncut. if you have a 3090, you can run this right now. free. no API. no subscription. opus structured reasoning on localhost. octopus invaders is next. same prompt that base qwen passed in 13 minutes and hermes 4.3 failed on 2x the hardware. i want to see if the distillation changes the outcome or just the style. more data soon.
1:53

Qwopus on a single RTX 3090. Claude Opus 4.6 reasoning distilled into Qwen 3.5 27B dense, running through Claude's own coding agent (claude code). 29-35 tok/s with thinking mode on. the jinja bug that kills thinking on base Qwen doesn't carry over. harness and model matched. the base model would pause mid task on Claude Code. just stop generating. that's why i ran it through OpenCode, which handles stalled states automatically. this distilled version doesn't stall. it waits for tool outputs, reads them, selfcorrects when something breaks, and keeps going. i gave it a benchmark analysis task. went 9 minutes autonomous. wrote a README nobody asked for. zero steering. video is 5x speed but fully uncut. if you have a 3090, you can run this right now. free. no API. no subscription. opus structured reasoning on localhost. octopus invaders is next. same prompt that base qwen passed in 13 minutes and hermes 4.3 failed on 2x the hardware. i want to see if the distillation changes the outcome or just the style. more data soon.

Sudo su

295,349 Aufrufe • vor 6 Monaten

Someone bought a Mac Mini for local AI. 64GB of unified memory seemed like the obvious choice. More memory. Bigger models. No VRAM limits. Then they put it against an RTX 3060. And things got weird. The Mac Mini M4 Pro pushed Qwen 2.5 at 38 tok/s. The old RTX 3060 with just 12GB VRAM hit 52 tok/s. That’s the interesting part about local AI. Apple and NVIDIA are solving different problems. Apple gives you a huge pool of memory. NVIDIA gives you dedicated compute and CUDA. One helps you fit bigger models. The other can run smaller models faster. Memory gets the model in. Compute gets the tokens out.
1:11

Someone bought a Mac Mini for local AI. 64GB of unified memory seemed like the obvious choice. More memory. Bigger models. No VRAM limits. Then they put it against an RTX 3060. And things got weird. The Mac Mini M4 Pro pushed Qwen 2.5 at 38 tok/s. The old RTX 3060 with just 12GB VRAM hit 52 tok/s. That’s the interesting part about local AI. Apple and NVIDIA are solving different problems. Apple gives you a huge pool of memory. NVIDIA gives you dedicated compute and CUDA. One helps you fit bigger models. The other can run smaller models faster. Memory gets the model in. Compute gets the tokens out.

KyzoroX

26,269 Aufrufe • vor 20 Tagen

the tiebreaker is done. qwen 3.5 27B dense. single RTX 3090. one prompt. zero steering. zero human edits. 1,827 lines across 10 files. 13 minutes. full thinking mode. runs on first load. hermes 4.3 got the same prompt with 2x 3090s and 5x the context it needed. wrote 1,249 lines, left empty files, needed 3 interventions, game was broken on load. same architecture class. same quant. hermes got double the hardware. completely different result. dense wasn't the problem. hermes was. but here's what got me. this model thinks at 27 tok/s. every single token carries 27 billion parameters of reasoning. MoE hit 112 tok/s but only 3B active per token. the dense model is slower and it doesn't matter. watch 13 minutes of autonomous coding on a consumer GPU with zero intervention and tell me speed is what matters. a year ago this wasn't possible. now it runs on hardware you can buy used for $900. no API. no subscription. no cloud. just a 3090 doing what data centers did 18 months ago. full unedited session in the video. every token, every file, every thinking chain. 16 minutes. hit play.
16:04

the tiebreaker is done. qwen 3.5 27B dense. single RTX 3090. one prompt. zero steering. zero human edits. 1,827 lines across 10 files. 13 minutes. full thinking mode. runs on first load. hermes 4.3 got the same prompt with 2x 3090s and 5x the context it needed. wrote 1,249 lines, left empty files, needed 3 interventions, game was broken on load. same architecture class. same quant. hermes got double the hardware. completely different result. dense wasn't the problem. hermes was. but here's what got me. this model thinks at 27 tok/s. every single token carries 27 billion parameters of reasoning. MoE hit 112 tok/s but only 3B active per token. the dense model is slower and it doesn't matter. watch 13 minutes of autonomous coding on a consumer GPU with zero intervention and tell me speed is what matters. a year ago this wasn't possible. now it runs on hardware you can buy used for $900. no API. no subscription. no cloud. just a 3090 doing what data centers did 18 months ago. full unedited session in the video. every token, every file, every thinking chain. 16 minutes. hit play.

Sudo su

91,135 Aufrufe • vor 6 Monaten

AI will replace the chrome in the coming year 🤯 Meet Neo by Norton Neo Browser — the first AI-native browser that doesn’t just open tabs… it thinks with you. You’re not searching anymore. You’re collaborating. Neo reads what’s on your screen, summarizes it instantly, and tells you what to do next. It understands context — your tasks, your flow, your intent. No more bouncing between Google and ChatGPT. Neo merges them into one seamless workspace where you ask, explore, and act in real time. It plugs into your apps too — turns your emails into to-dos, meetings into reminders, research into insight. And because it’s built by Norton, the global cybersecurity giant, it’s smarter, safer, and ad-free. This isn’t just a browser. It’s your intelligent co-pilot for the internet. Welcome to the future of browsing. Smarter. Safer. Simpler. 👉 Join the waitlist:
2:05

AI will replace the chrome in the coming year 🤯 Meet Neo by Norton Neo Browser — the first AI-native browser that doesn’t just open tabs… it thinks with you. You’re not searching anymore. You’re collaborating. Neo reads what’s on your screen, summarizes it instantly, and tells you what to do next. It understands context — your tasks, your flow, your intent. No more bouncing between Google and ChatGPT. Neo merges them into one seamless workspace where you ask, explore, and act in real time. It plugs into your apps too — turns your emails into to-dos, meetings into reminders, research into insight. And because it’s built by Norton, the global cybersecurity giant, it’s smarter, safer, and ad-free. This isn’t just a browser. It’s your intelligent co-pilot for the internet. Welcome to the future of browsing. Smarter. Safer. Simpler. 👉 Join the waitlist:

Hamza Khalid

90,523 Aufrufe • vor 11 Monaten

Meet WebBrain: An Open-Source, Local-First AI Browser Agent That Reads Pages and Automates Tasks in Chrome and Firefox WebBrain lives inside your browser and can run entirely on your own local model — no cloud, no account, no data leaving your machine. Most "AI browser agents" are a chat box that pastes your page into someone else's server. That's not an agent that lives where you browse — and WebBrain draws a very clear line between the two. It's an open-source (MIT), local-first browser agent for Chrome and Firefox. It runs inside your existing authenticated session, on a model you pick — so with llama.cpp or Ollama, nothing leaves your machine. Here's what's actually interesting: → Two modes, cleanly separated. Ask reads the page (read-only, content scripts). Act clicks and types through the Chrome DevTools Protocol (chrome.debugger) — trusted input events that modern sites honor, reaching cross-origin iframes and shadow DOM. → UI-first by design. For anything that submits, sends, or buys, it drives the visible UI and refuses to hit REST/GraphQL endpoints directly. It starts read-only and asks before consequential actions. → Bring any model. llama.cpp, Ollama, LM Studio, vLLM — or OpenAI, Claude, Gemini, DeepSeek, Groq, OpenRouter. Recommended local: Qwen 3.6 35B (Qwen3.6-35B-A3B), which beat Gemma 4 on the project's screenshot benchmark. → Tuned for cost and privacy. Token-conscious screenshots, oldest-first context trimming, a dedicated vision model, 40+ tools (~20 in Compact mode). No telemetry. No accounts. Full analysis: GitHub Repo: Chrome Extension: Firefox Add-on: Portal:
0:47

Meet WebBrain: An Open-Source, Local-First AI Browser Agent That Reads Pages and Automates Tasks in Chrome and Firefox WebBrain lives inside your browser and can run entirely on your own local model — no cloud, no account, no data leaving your machine. Most "AI browser agents" are a chat box that pastes your page into someone else's server. That's not an agent that lives where you browse — and WebBrain draws a very clear line between the two. It's an open-source (MIT), local-first browser agent for Chrome and Firefox. It runs inside your existing authenticated session, on a model you pick — so with llama.cpp or Ollama, nothing leaves your machine. Here's what's actually interesting: → Two modes, cleanly separated. Ask reads the page (read-only, content scripts). Act clicks and types through the Chrome DevTools Protocol (chrome.debugger) — trusted input events that modern sites honor, reaching cross-origin iframes and shadow DOM. → UI-first by design. For anything that submits, sends, or buys, it drives the visible UI and refuses to hit REST/GraphQL endpoints directly. It starts read-only and asks before consequential actions. → Bring any model. llama.cpp, Ollama, LM Studio, vLLM — or OpenAI, Claude, Gemini, DeepSeek, Groq, OpenRouter. Recommended local: Qwen 3.6 35B (Qwen3.6-35B-A3B), which beat Gemma 4 on the project's screenshot benchmark. → Tuned for cost and privacy. Token-conscious screenshots, oldest-first context trimming, a dedicated vision model, 40+ tools (~20 in Compact mode). No telemetry. No accounts. Full analysis: GitHub Repo: Chrome Extension: Firefox Add-on: Portal:

Marktechpost AI

203,011 Aufrufe • vor 2 Monaten

first impressions of qwen 3.5 27B dense on a single RTX 3090. 35 tok/s. from 4K all the way to 300K+ context. no speed drop. hermes 4.3 started at 35 and degraded to 15 as context filled. qwen dense holds. MoE held 112 flat. 3x faster but only 3B of 35B active per token. architecture tradeoff. Q4_K_M on 16.7GB. native context 262K. pushed past training limit to 376K before VRAM ceiling on 24GB. tried q8 KV cache at 262K, speed collapsed to 11 tok/s. q4_0 KV is the sweet spot. flash attention mandatory. built in reasoning mode. the model thinks step by step before it answers. full chain of thought surviving Q4 quant. 1,799+ token thinking chains with self correction loops. on a single consumer GPU. gave it one prompt: "build a realtime particle galaxy simulation in one HTML file." 3,340 tokens. 95 seconds. one shot. ran on first load. full reasoning and coding in the video below. optimal config if you want to skip the hours of testing: llama-server -ngl 99 -c 262144 -fa on --cache-type-k q4_0 --cache-type-v q4_0 this is just the warmup. octopus invaders is next: 10 files, 3,400+ lines, zero steering. the prompt hermes quit at 22%. already more impressed than expected. full results coming soon.
2:02

first impressions of qwen 3.5 27B dense on a single RTX 3090. 35 tok/s. from 4K all the way to 300K+ context. no speed drop. hermes 4.3 started at 35 and degraded to 15 as context filled. qwen dense holds. MoE held 112 flat. 3x faster but only 3B of 35B active per token. architecture tradeoff. Q4_K_M on 16.7GB. native context 262K. pushed past training limit to 376K before VRAM ceiling on 24GB. tried q8 KV cache at 262K, speed collapsed to 11 tok/s. q4_0 KV is the sweet spot. flash attention mandatory. built in reasoning mode. the model thinks step by step before it answers. full chain of thought surviving Q4 quant. 1,799+ token thinking chains with self correction loops. on a single consumer GPU. gave it one prompt: "build a realtime particle galaxy simulation in one HTML file." 3,340 tokens. 95 seconds. one shot. ran on first load. full reasoning and coding in the video below. optimal config if you want to skip the hours of testing: llama-server -ngl 99 -c 262144 -fa on --cache-type-k q4_0 --cache-type-v q4_0 this is just the warmup. octopus invaders is next: 10 files, 3,400+ lines, zero steering. the prompt hermes quit at 22%. already more impressed than expected. full results coming soon.

Sudo su

122,051 Aufrufe • vor 6 Monaten