Sudo su's banner

Sudo su

@sudoingX • 39,554 subscribers

GPU/local LLM. more RAM and OSS... everywhere

Shorts

hey if you have a 3060, or any GPU with 8GB or more sitting in a drawer right now, that thing can run 9 billion parameters of intelligence autonomously. and you don't know it yet. 2 hours ago i posted that 9B hit a ceiling. 2,699 lines across 11 files. blank screen. said the limit for autonomous multifile coding on 9 billion parameters is real. then i audited every file. found 11 bugs. exact file, exact line, exact fix. duplicate variable declarations killing the script loader. a canvas reference never connected to the DOM. enemies with no movement logic. particle systems called on the class instead of the instance. fed that list as a single prompt to the same Qwen 3.5 9B on the same RTX 3060 through Hermes Agent. it fixed all 11. surgically. patch level edits across 4 files. no rewrites. no hallucinated changes. game boots. enemies spawn, move, collide. background renders. particles fire. and here's what nobody is talking about. this is a 9 billion parameter model running a full agentic framework. Hermes Agent with 31 tools. file operations, terminal, browser, code execution. not a single tool call failed. the agent chain never broke. most people think you need 70B+ for reliable tool use. this is 9B on 12 gigs doing it clean. the model didn't fail. my prompting strategy did. the ceiling is not the parameter count. the ceiling is how you prompt it. this is not done. bullets don't fire yet. boss fights need wiring. but the screen that was black 2 hours ago now has a full game rendering in real time. iterating right now. anyone with a GPU from the last 5 years should be paying attention to what is happening right now.

hey if you have a 3060, or any GPU with 8GB or more sitting in a drawer right now, that thing can run 9 billion parameters of intelligence autonomously. and you don't know it yet. 2 hours ago i posted that 9B hit a ceiling. 2,699 lines across 11 files. blank screen. said the limit for autonomous multifile coding on 9 billion parameters is real. then i audited every file. found 11 bugs. exact file, exact line, exact fix. duplicate variable declarations killing the script loader. a canvas reference never connected to the DOM. enemies with no movement logic. particle systems called on the class instead of the instance. fed that list as a single prompt to the same Qwen 3.5 9B on the same RTX 3060 through Hermes Agent. it fixed all 11. surgically. patch level edits across 4 files. no rewrites. no hallucinated changes. game boots. enemies spawn, move, collide. background renders. particles fire. and here's what nobody is talking about. this is a 9 billion parameter model running a full agentic framework. Hermes Agent with 31 tools. file operations, terminal, browser, code execution. not a single tool call failed. the agent chain never broke. most people think you need 70B+ for reliable tool use. this is 9B on 12 gigs doing it clean. the model didn't fail. my prompting strategy did. the ceiling is not the parameter count. the ceiling is how you prompt it. this is not done. bullets don't fire yet. boss fights need wiring. but the screen that was black 2 hours ago now has a full game rendering in real time. iterating right now. anyone with a GPU from the last 5 years should be paying attention to what is happening right now.

684,336 views

single RTX 3090. 24 GB VRAM. Qwen3.5-35B-A3B. 4-bit quant, 113 tokens per second at full 262K context harnessing Claude Code locally with no API, no subscription, no proxy. told it what it is. 30 Mamba2 layers, 10 attention, 256 experts, 8 active per token. said "build something that shows off what you can do." it visualized its own architecture. interactive. tokens flowing through layers. 256 experts lighting up on routing. served in the browser from the same GPU running inference. single prompt. then i said level up. 3D. Three.js. separate files. flythrough camera. clickable layers. it planned first, scaffolded 6 files, hit one API bug, fixed it itself, then optimized for smooth framerate. two iterations to a working 3D neural network explorer. llama.cpp just merged a native Anthropic endpoint. Claude Code points at localhost. the whole setup is two commands. no LiteLLM. no proxy config. the open source models coming out of china right now are genuinely changing what's possible on consumer hardware. respect to the Qwen team. this is acceleration.

single RTX 3090. 24 GB VRAM. Qwen3.5-35B-A3B. 4-bit quant, 113 tokens per second at full 262K context harnessing Claude Code locally with no API, no subscription, no proxy. told it what it is. 30 Mamba2 layers, 10 attention, 256 experts, 8 active per token. said "build something that shows off what you can do." it visualized its own architecture. interactive. tokens flowing through layers. 256 experts lighting up on routing. served in the browser from the same GPU running inference. single prompt. then i said level up. 3D. Three.js. separate files. flythrough camera. clickable layers. it planned first, scaffolded 6 files, hit one API bug, fixed it itself, then optimized for smooth framerate. two iterations to a working 3D neural network explorer. llama.cpp just merged a native Anthropic endpoint. Claude Code points at localhost. the whole setup is two commands. no LiteLLM. no proxy config. the open source models coming out of china right now are genuinely changing what's possible on consumer hardware. respect to the Qwen team. this is acceleration.

110,464 views

this is the worst local AI will ever be. tomorrow it gets faster. next month the models get smarter. next year your GPU runs what a data center runs today. Qwen3.5-35B-A3B on a single 3090. told it to visualize its own expert routing. 256 experts, 8 active per token, rendered in 3D on the same GPU running inference. no API key. no subscription. no permission needed. closed AI isn't losing ground. it's losing the argument.

this is the worst local AI will ever be. tomorrow it gets faster. next month the models get smarter. next year your GPU runs what a data center runs today. Qwen3.5-35B-A3B on a single 3090. told it to visualize its own expert routing. 256 experts, 8 active per token, rendered in 3D on the same GPU running inference. no API key. no subscription. no permission needed. closed AI isn't losing ground. it's losing the argument.

106,822 views

there is so much real data just sitting in the open right now it's almost funny. four years of starlight on every star, a NASA archive that's been free for over a decade, detectors still recording the sky tonight, and barely anyone has a net pointed at any of it. so i pointed one. this is me pulling the planet data, the data loading is the boring part. the net i built to read it, the wall it hit, and what that taught me about where AI goes next, that's the full story, and it drops tonight. the data's public, the tools are free, the box fits on a desk. what's stopping you. you can just do things anon.

there is so much real data just sitting in the open right now it's almost funny. four years of starlight on every star, a NASA archive that's been free for over a decade, detectors still recording the sky tonight, and barely anyone has a net pointed at any of it. so i pointed one. this is me pulling the planet data, the data loading is the boring part. the net i built to read it, the wall it hit, and what that taught me about where AI goes next, that's the full story, and it drops tonight. the data's public, the tools are free, the box fits on a desk. what's stopping you. you can just do things anon.

60,445 views

i gave bonsai, 3.9 gb model one spec to build a working rotary engine autonomously, using hermes agent. all on rtx 3060ti 8gb. watch it go from a black canvas to a running wankel. the epitrochoid housing, a triangular rotor turning at exactly a third of shaft speed, three chambers firing intake, compression, power, exhaust, and ten geometry tests it had to pass to count as done. a provably correct simulation, not a drawing. then i left it alone and it worked. wrote the code, ran its own tests, hit failures, read them, fixed its own lines, until the engine spun and all ten went green. a real agent loop, and i watched every step. it broke three times getting here and clawed back every one. that honest part is in the reply. and this is a sneak. i've been building a show around runs exactly like this, and it's almost ready. more very soon. just getting started.

i gave bonsai, 3.9 gb model one spec to build a working rotary engine autonomously, using hermes agent. all on rtx 3060ti 8gb. watch it go from a black canvas to a running wankel. the epitrochoid housing, a triangular rotor turning at exactly a third of shaft speed, three chambers firing intake, compression, power, exhaust, and ten geometry tests it had to pass to count as done. a provably correct simulation, not a drawing. then i left it alone and it worked. wrote the code, ran its own tests, hit failures, read them, fixed its own lines, until the engine spun and all ten went green. a real agent loop, and i watched every step. it broke three times getting here and clawed back every one. that honest part is in the reply. and this is a sneak. i've been building a show around runs exactly like this, and it's almost ready. more very soon. just getting started.

35,069 views

let me save you 3 hours of head scratching. if you're running local models like Qwen3.5-35B-A3B through Claude Code via llama.cpp's Anthropic endpoint, the chain will break every 3 to 5 minutes. tool call fails. flow stops. you reprompt. it recovers. 2 minutes later it stops again. the model is fine. the harness chokes on local inference latency. switch to OpenCode. same localhost endpoint. same model. same GPU. the chain doesn't break. the tradeoff: OpenCode sometimes loops. the model forgets what it already read and repeats the same tool call. but a loop you can interrupt. a broken chain kills your momentum and you start over. watch both side by side. proprietary agent vs open source agent. same 3B model. different failure modes. pick your poison.

let me save you 3 hours of head scratching. if you're running local models like Qwen3.5-35B-A3B through Claude Code via llama.cpp's Anthropic endpoint, the chain will break every 3 to 5 minutes. tool call fails. flow stops. you reprompt. it recovers. 2 minutes later it stops again. the model is fine. the harness chokes on local inference latency. switch to OpenCode. same localhost endpoint. same model. same GPU. the chain doesn't break. the tradeoff: OpenCode sometimes loops. the model forgets what it already read and repeats the same tool call. but a loop you can interrupt. a broken chain kills your momentum and you start over. watch both side by side. proprietary agent vs open source agent. same 3B model. different failure modes. pick your poison.

72,554 views

this is the worst local ai will ever be. it only gets better from here. if you are not expanding your mind with these small models you are missing what's happening right now 99 percent tool call success rate. when steered well with the right skills and a framework like hermes agent the node becomes a cognition layer. not a chatbot. not a toy. an extension of how you think. i was cranking this node at 35 to 50 tok/s all day on personal experiments and now after all the work is done qwen 3.5 9B is iterating on its own code. the game it created. fixing its own bugs autonomously. and the part you should probably not miss is that all of this is happening on a RTX 3060. not an H100. not an A100. the card most of you have sitting in a drawer right now. if you just open that drawer and put that intelligence to work every tensor core on that card should be running for you. your work. your experiments. your thinking. you all have it but because nobody told you what this hardware can actually do in 2026 you never tried. the day it unlocks is the day you test your workload, understand the tradeoffs, debug the loops, and then decide if you need to scale the hardware. there is no point buying 3 mac studios when things done well you can squeeze a similar level of intelligence from 9B compared to 70B. but only when you create the right environment for your model through the right harness. and let me tell you i have tried claude code as a local harness. i have tried opencode. i have tried various others. somehow i landed on hermes agent and never left. there is something magical going on at Nous Research. the tool call parsers, the skills system, the way it handles small models natively. nothing else comes close for local inference. own your cognition. your AI. your agent. your prompts. your experiments. why give them away for free. those are who you are and they don't belong on someone else's servers being monitored. just give it a shot with your existing hardware. you run into a problem the community will help you. and if you are migrating from openclaw to hermes i will personally help you make the switch.

this is the worst local ai will ever be. it only gets better from here. if you are not expanding your mind with these small models you are missing what's happening right now 99 percent tool call success rate. when steered well with the right skills and a framework like hermes agent the node becomes a cognition layer. not a chatbot. not a toy. an extension of how you think. i was cranking this node at 35 to 50 tok/s all day on personal experiments and now after all the work is done qwen 3.5 9B is iterating on its own code. the game it created. fixing its own bugs autonomously. and the part you should probably not miss is that all of this is happening on a RTX 3060. not an H100. not an A100. the card most of you have sitting in a drawer right now. if you just open that drawer and put that intelligence to work every tensor core on that card should be running for you. your work. your experiments. your thinking. you all have it but because nobody told you what this hardware can actually do in 2026 you never tried. the day it unlocks is the day you test your workload, understand the tradeoffs, debug the loops, and then decide if you need to scale the hardware. there is no point buying 3 mac studios when things done well you can squeeze a similar level of intelligence from 9B compared to 70B. but only when you create the right environment for your model through the right harness. and let me tell you i have tried claude code as a local harness. i have tried opencode. i have tried various others. somehow i landed on hermes agent and never left. there is something magical going on at Nous Research. the tool call parsers, the skills system, the way it handles small models natively. nothing else comes close for local inference. own your cognition. your AI. your agent. your prompts. your experiments. why give them away for free. those are who you are and they don't belong on someone else's servers being monitored. just give it a shot with your existing hardware. you run into a problem the community will help you. and if you are migrating from openclaw to hermes i will personally help you make the switch.

58,717 views

watch anon. a 27b model thinking out loud on gtx 1660 super, i asked bonsai what model it is and it reasoned through the whole answer before it spoke. 20 tokens a second, gpu pinned at 125 watts, 4.25 of 6 gigs used. bonsai is a 1bit quant of qwen 3.6 27b, the king of the 3090 crushed to 3.5gb. turns out the hardware was in your drawer the whole time. you just needed the quant to catch up.

watch anon. a 27b model thinking out loud on gtx 1660 super, i asked bonsai what model it is and it reasoned through the whole answer before it spoke. 20 tokens a second, gpu pinned at 125 watts, 4.25 of 6 gigs used. bonsai is a 1bit quant of qwen 3.6 27b, the king of the 3090 crushed to 3.5gb. turns out the hardware was in your drawer the whole time. you just needed the quant to catch up.

26,442 views

look what a single consumer GPU just built. gave Qwen3.5-35B-A3B one prompt: build a cloud GPU marketplace with pricing cards, deploy templates, and a benchmark leaderboard. it planned the layout, wrote the animations, populated the data, and served it. one shot. one HTML file. then i told it to iterate. split the hero, add a floating GPU with neural network animation. glassmorphism on the cards. done. done. done. three rounds, no confusion, no regressions. 4-bit quantized. 19.7 GB. single RTX 3090. full coding agent claude code harness running on localhost. no API calls leaving my machine. no subscription. no rate limits. earlier today i pointed it at my own production website. it curled the HTML, found every broken link, and told me "pretty shell, empty core. would not recommend." then built a better version from scratch. local inference stops being a demo when you actually steer it. the models are there. they understand intent. but you have to meet them halfway with good prompts, clear context, and real project structure. that's the skill gap now. not the models. the steering. more experiments coming. i genuinely cannot stop playing with this thing.

look what a single consumer GPU just built. gave Qwen3.5-35B-A3B one prompt: build a cloud GPU marketplace with pricing cards, deploy templates, and a benchmark leaderboard. it planned the layout, wrote the animations, populated the data, and served it. one shot. one HTML file. then i told it to iterate. split the hero, add a floating GPU with neural network animation. glassmorphism on the cards. done. done. done. three rounds, no confusion, no regressions. 4-bit quantized. 19.7 GB. single RTX 3090. full coding agent claude code harness running on localhost. no API calls leaving my machine. no subscription. no rate limits. earlier today i pointed it at my own production website. it curled the HTML, found every broken link, and told me "pretty shell, empty core. would not recommend." then built a better version from scratch. local inference stops being a demo when you actually steer it. the models are there. they understand intent. but you have to meet them halfway with good prompts, clear context, and real project structure. that's the skill gap now. not the models. the steering. more experiments coming. i genuinely cannot stop playing with this thing.

37,373 views

here's how the whole thing works. claude code doesn't care what's behind the API. it just sends requests and expects responses. so i pointed it at my own machine instead of anthropic's servers. llama-server runs the model locally. LiteLLM sits in between and translates the API format. claude code thinks it's talking to claude. it's talking to qwen on localhost. the setup: 2x 3090s, 38 layers on GPU, 10 on CPU. 128K context window. generation is only 7 tok/s but the tradeoff is worth it. 128K means the agent can hold an entire project in memory without losing context midtask. claude code alone loads a 17.5K token system prompt on every request. tool definitions, safety rules, agent behavior. that's your baseline before you even say hello. pushed as far as i could tonight. what surprised me most wasn't the speed. it was the iteration quality. first prompt gave me a working particle sim. second prompt, the model read its own 564 lines, understood the architecture, and added trails, explosions, gravity wells, bloom effects. no handholding. 4bit quantized. 45GB on two consumer cards. running a full coding agent autonomously. detailed article coming. full benchmarks, hardware breakdowns, engine debugging, code quality. everything from setup to what broke and why.

here's how the whole thing works. claude code doesn't care what's behind the API. it just sends requests and expects responses. so i pointed it at my own machine instead of anthropic's servers. llama-server runs the model locally. LiteLLM sits in between and translates the API format. claude code thinks it's talking to claude. it's talking to qwen on localhost. the setup: 2x 3090s, 38 layers on GPU, 10 on CPU. 128K context window. generation is only 7 tok/s but the tradeoff is worth it. 128K means the agent can hold an entire project in memory without losing context midtask. claude code alone loads a 17.5K token system prompt on every request. tool definitions, safety rules, agent behavior. that's your baseline before you even say hello. pushed as far as i could tonight. what surprised me most wasn't the speed. it was the iteration quality. first prompt gave me a working particle sim. second prompt, the model read its own 564 lines, understood the architecture, and added trails, explosions, gravity wells, bloom effects. no handholding. 4bit quantized. 45GB on two consumer cards. running a full coding agent autonomously. detailed article coming. full benchmarks, hardware breakdowns, engine debugging, code quality. everything from setup to what broke and why.

37,623 views

5 days ago it took 2 GPUs to build this. today it takes 1. same prompt. same particle simulation. completely different model. Qwen-Coder-Next (80B) on 2x 3090s. 46 tok/s. 564 lines. 2 iterations to get it working. 48GB VRAM across two cards just to hold it. Qwen3.5-35B-A3B on a single 3090. 112 tok/s. 461 lines. first try. cleaner code, fewer lines, better structured. 19.7GB on disk with 4GB VRAM to spare. half the parameters. one GPU instead of two. 2.4x faster. and the output actually improved. this is what happens when architecture catches up to ambition. Gated Delta Networks(Mamba2 variant) hybrid with sparse MoE. 3B active params out of 35B per token. efficiency at the architecture level, not just quantization. the curve isn't flattening. it's steepening.

5 days ago it took 2 GPUs to build this. today it takes 1. same prompt. same particle simulation. completely different model. Qwen-Coder-Next (80B) on 2x 3090s. 46 tok/s. 564 lines. 2 iterations to get it working. 48GB VRAM across two cards just to hold it. Qwen3.5-35B-A3B on a single 3090. 112 tok/s. 461 lines. first try. cleaner code, fewer lines, better structured. 19.7GB on disk with 4GB VRAM to spare. half the parameters. one GPU instead of two. 2.4x faster. and the output actually improved. this is what happens when architecture catches up to ambition. Gated Delta Networks(Mamba2 variant) hybrid with sparse MoE. 3B active params out of 35B per token. efficiency at the architecture level, not just quantization. the curve isn't flattening. it's steepening.

34,645 views

the 24gb vram tier is enough for most builder work in 2026. gemma 4 31b dense on my rog scar 18 just autonomously built a production hero section in one prompt, one html file and 5 minutes end to end. hardware: rog scar 18, rtx 5090 laptop 24gb vram. model: google gemma 4 31b dense at q4_k_m quant, using 22.8 of 24gb. engine: llama.cpp built for blackwell (sm_120). harness: hermes agent with native tool parsing. speed: 15 tok/s sustained, 94 watts, 50c. flags i used: ./build/bin/llama-server -m ~/models/gemma4-31b/google_gemma-4-31B-it-Q4_K_M.gguf -ngl 99 -c 131072 -np 1 -fa on --cache-type-k q4_0 --cache-type-v q4_0 --jinja --host 127.0.0.1 --port 8080 if you own 24gb vram in 2026, you have enough for most ui work, most agentic coding, most autonomous builds. no subscription, no one logging your prompts. a dense open model on consumer hardware shipping real software on your desk. this was the warmup. full page next on same hardware, then the octopus invaders final multifile autonomous challenge.

the 24gb vram tier is enough for most builder work in 2026. gemma 4 31b dense on my rog scar 18 just autonomously built a production hero section in one prompt, one html file and 5 minutes end to end. hardware: rog scar 18, rtx 5090 laptop 24gb vram. model: google gemma 4 31b dense at q4_k_m quant, using 22.8 of 24gb. engine: llama.cpp built for blackwell (sm_120). harness: hermes agent with native tool parsing. speed: 15 tok/s sustained, 94 watts, 50c. flags i used: ./build/bin/llama-server -m ~/models/gemma4-31b/google_gemma-4-31B-it-Q4_K_M.gguf -ngl 99 -c 131072 -np 1 -fa on --cache-type-k q4_0 --cache-type-v q4_0 --jinja --host 127.0.0.1 --port 8080 if you own 24gb vram in 2026, you have enough for most ui work, most agentic coding, most autonomous builds. no subscription, no one logging your prompts. a dense open model on consumer hardware shipping real software on your desk. this was the warmup. full page next on same hardware, then the octopus invaders final multifile autonomous challenge.

19,576 views

most of you don't know how hard hermes agent is optimized for local AI at the system level. watch the full setup flow on screen. you paste an openai-compatible v1 endpoint, hermes auto-detects every model running behind it. doesn't matter if it's llama.cpp or vllm or any compatible server, all your models surface and become selectable in seconds. no config gymnastics, no manual model list. then it goes deeper. hermes ships with per model parsers, prompt template auto-handling, tool call format detection per model architecture, thinking mode awareness, all the small friction points other harnesses leak on. these were not built for cloud apis with one canonical model. they were built for builders running 10 different local models across 10 different stacks. cloud first harnesses bolt local support on top. hermes agent is local first from the architecture out. that's the system level gap. if you're getting started on local AI, this is the harness you start with. try for yourself and find out. anyone serious about local AI lands here eventually.

most of you don't know how hard hermes agent is optimized for local AI at the system level. watch the full setup flow on screen. you paste an openai-compatible v1 endpoint, hermes auto-detects every model running behind it. doesn't matter if it's llama.cpp or vllm or any compatible server, all your models surface and become selectable in seconds. no config gymnastics, no manual model list. then it goes deeper. hermes ships with per model parsers, prompt template auto-handling, tool call format detection per model architecture, thinking mode awareness, all the small friction points other harnesses leak on. these were not built for cloud apis with one canonical model. they were built for builders running 10 different local models across 10 different stacks. cloud first harnesses bolt local support on top. hermes agent is local first from the architecture out. that's the system level gap. if you're getting started on local AI, this is the harness you start with. try for yourself and find out. anyone serious about local AI lands here eventually.

14,407 views

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

when you realize the 6gb uncensored model on your own gpu just did the work opus 5.5 refused

when you realize the 6gb uncensored model on your own gpu just did the work opus 5.5 refused

188,824 views • 3 days ago

when you realize your 12gb gaming gpu runs a 27b ai model at 50 tok/s and does overnight agentic tasks

when you realize your 12gb gaming gpu runs a 27b ai model at 50 tok/s and does overnight agentic tasks

233,636 views • 24 days ago

this is what 12gb of vram builds in 2026, absolute magic > rtx 3060 12gb, #1 gpu on steam > bonsai 2 27b + mtp, 5.95 gb of weights > hermes agent, 5 hours, 328k tokens written > 8 js files, 2,368 lines, zero hand written code > 50 tok/s fresh, 22 tok/s average, 125k context watch the full video, 5 hours in 12 minutes of pure dance of a local ai model on rtx 3060 12gb vram, and stay till the end for the full gameplay. this entire game was built by bonsai2, a qwen 3.8 27b dense compressed to ternary, and this small model is punching way above its weight. it built multi file engineering work using hermes agent, sure it's not fast but perfect for overnights and routine work and the quality is insane, and context holding is another best one, it does not lose the thread. i ran PrismML bonsai 1 made from qwen 3.6 27b dense and i built things with it, but this time with the latest base model qwen 3.8 these results are insane, and because i loved building with it a lot i thought many more of you would run it because this gpu exists in almost every home. so i packaged mtp, doubled the speed from 26 tok/s to 50 tok/s, packed the prefill fix in and released it on huggingface, almost 4,000 downloads in 3 days. i'll leave a link below.

this is what 12gb of vram builds in 2026, absolute magic > rtx 3060 12gb, #1 gpu on steam > bonsai 2 27b + mtp, 5.95 gb of weights > hermes agent, 5 hours, 328k tokens written > 8 js files, 2,368 lines, zero hand written code > 50 tok/s fresh, 22 tok/s average, 125k context watch the full video, 5 hours in 12 minutes of pure dance of a local ai model on rtx 3060 12gb vram, and stay till the end for the full gameplay. this entire game was built by bonsai2, a qwen 3.8 27b dense compressed to ternary, and this small model is punching way above its weight. it built multi file engineering work using hermes agent, sure it's not fast but perfect for overnights and routine work and the quality is insane, and context holding is another best one, it does not lose the thread. i ran PrismML bonsai 1 made from qwen 3.6 27b dense and i built things with it, but this time with the latest base model qwen 3.8 these results are insane, and because i loved building with it a lot i thought many more of you would run it because this gpu exists in almost every home. so i packaged mtp, doubled the speed from 26 tok/s to 50 tok/s, packed the prefill fix in and released it on huggingface, almost 4,000 downloads in 3 days. i'll leave a link below.

87,876 views • 24 days ago

single most useful thing for my local ai setup that made life easy is my 2x DGX Spark becoming one serving box for every device i own here is how i do it: > 1. the two sparks run one vLLM server together with one endpoint, i use dgx sparks, you can use any nodes that can run an llm > 2. install tailscale to have all your nodes and machines on the same tailnet, the endpoint lives there so every device reaches it from anywhere and no port is open to the internet > 3. that one tailnet endpoint with the exposed port is your base url for everything, it speaks OpenAI chat, OpenAI Responses and Anthropic Messages, so any chat app, coding agent or phone bot you point at it talks to the same model > 4. the server also answers to the name "local", so clients can ask for "local" instead of a model name and when i swap the model on the servers nothing on the devices needs a config change. right now it's serving GLM 5.3-Flash, next week it can be something else this video below is the whole thing running end to end, orange dots are requests going in, white and green are tokens coming back it takes however many requests you throw at it, your config decides how many run at once and the rest wait their turn, when several run together each request's tok/s drops a bit while the box moves more tokens in total. mine is set to one at a time right now, so when every device fires together they queue up this setup really improved my quality of life with local ai, if you get stuck anywhere setting it up leave a comment and i'll help you

single most useful thing for my local ai setup that made life easy is my 2x DGX Spark becoming one serving box for every device i own here is how i do it: > 1. the two sparks run one vLLM server together with one endpoint, i use dgx sparks, you can use any nodes that can run an llm > 2. install tailscale to have all your nodes and machines on the same tailnet, the endpoint lives there so every device reaches it from anywhere and no port is open to the internet > 3. that one tailnet endpoint with the exposed port is your base url for everything, it speaks OpenAI chat, OpenAI Responses and Anthropic Messages, so any chat app, coding agent or phone bot you point at it talks to the same model > 4. the server also answers to the name "local", so clients can ask for "local" instead of a model name and when i swap the model on the servers nothing on the devices needs a config change. right now it's serving GLM 5.3-Flash, next week it can be something else this video below is the whole thing running end to end, orange dots are requests going in, white and green are tokens coming back it takes however many requests you throw at it, your config decides how many run at once and the rest wait their turn, when several run together each request's tok/s drops a bit while the box moves more tokens in total. mine is set to one at a time right now, so when every device fires together they queue up this setup really improved my quality of life with local ai, if you get stuck anywhere setting it up leave a comment and i'll help you

32,890 views • 17 days ago

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.

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.

296,392 views • 7 months ago

the timeline spent two days saying bonsai 2 cannot build. here is 1 hour 24 minutes of it building, one shot, from one paragraph, on an rtx 3060 12gb, sped to 8x so you can watch the whole thing. what you are watching is a 5.9gb ternary compression of qwen 3.8 27b, served through the prismml llama.cpp fork with the full 262k window resident, writing a terminal gpu monitor in one html file. it thinks, it calls a tool mid build, it keeps going until finish. the clock in the corner is real. 8x, one frozen recorder stretch cut, nothing else touched. for anyone new here, the 3060 is the most owned gpu on steam, a five year old card, and this is the kind of card the model was compressed for. for anyone who followed july, bonsai 1 built the wankel engine on an 8gb 3060 ti, this is the sequel one tier up. the numbers, live server, thinking on > 26 tok/s fresh, 22 at 12k, 18 at 35k, 13 tok/s at 77k deep > 262k context resident on 11.7 of 12 gb > 76k tokens, 77 minutes, 149 of 150 w the entire run if you own a 12gb card, this runs on it tonight. serve line, model, fork and the exact prompt in the replies.

the timeline spent two days saying bonsai 2 cannot build. here is 1 hour 24 minutes of it building, one shot, from one paragraph, on an rtx 3060 12gb, sped to 8x so you can watch the whole thing. what you are watching is a 5.9gb ternary compression of qwen 3.8 27b, served through the prismml llama.cpp fork with the full 262k window resident, writing a terminal gpu monitor in one html file. it thinks, it calls a tool mid build, it keeps going until finish. the clock in the corner is real. 8x, one frozen recorder stretch cut, nothing else touched. for anyone new here, the 3060 is the most owned gpu on steam, a five year old card, and this is the kind of card the model was compressed for. for anyone who followed july, bonsai 1 built the wankel engine on an 8gb 3060 ti, this is the sequel one tier up. the numbers, live server, thinking on > 26 tok/s fresh, 22 at 12k, 18 at 35k, 13 tok/s at 77k deep > 262k context resident on 11.7 of 12 gb > 76k tokens, 77 minutes, 149 of 150 w the entire run if you own a 12gb card, this runs on it tonight. serve line, model, fork and the exact prompt in the replies.

41,330 views • 29 days ago

i think spacex browser game just fixed my fried focus, lol spent 20 min last night manually docking a crew dragon to the iss and my head went dead quiet for the first time in weeks. it's the real spacex sim, free, runs in the browser, no install. ease the translation in, null your rotation rate, hold the crosshairs dead center, soft capture. and that's the actual manual procedure the crew falls back on if autonomous docking ever drops. go dock something tonight.

i think spacex browser game just fixed my fried focus, lol spent 20 min last night manually docking a crew dragon to the iss and my head went dead quiet for the first time in weeks. it's the real spacex sim, free, runs in the browser, no install. ease the translation in, null your rotation rate, hold the crosshairs dead center, soft capture. and that's the actual manual procedure the crew falls back on if autonomous docking ever drops. go dock something tonight.

155,016 views • 4 months ago

nvidia's 3B mamba destroyed alibaba's 3B deltanet on the same RTX 3090. only 24 days between releases. same active parameters, same VRAM tier, completely different architectures. nemotron cascade 2: 187 tok/s. flat from 4K to 625K context. zero speed loss. flags: -ngl 99 -np 1. that's it. no context flags, no KV cache tricks. auto-allocates 625K. qwen 3.5 35B-A3B: 112 tok/s. flat from 4K to 262K context. zero speed loss. flags: -ngl 99 -np 1 -c 262144 --cache-type-k q8_0 --cache-type-v q8_0. needed KV cache quantization to fit 262K. both models held a flat line across every context level. both architectures are context-independent. but nvidia's mamba2 is 67% faster at generating tokens on the exact same hardware and needs fewer flags to get there. same node, same GPU, same everything. the only variable is the model. gold medal math olympiad winner running at 187 tokens per second on single RTX 3090 a card from 6 years ago. nvidia cooked.

nvidia's 3B mamba destroyed alibaba's 3B deltanet on the same RTX 3090. only 24 days between releases. same active parameters, same VRAM tier, completely different architectures. nemotron cascade 2: 187 tok/s. flat from 4K to 625K context. zero speed loss. flags: -ngl 99 -np 1. that's it. no context flags, no KV cache tricks. auto-allocates 625K. qwen 3.5 35B-A3B: 112 tok/s. flat from 4K to 262K context. zero speed loss. flags: -ngl 99 -np 1 -c 262144 --cache-type-k q8_0 --cache-type-v q8_0. needed KV cache quantization to fit 262K. both models held a flat line across every context level. both architectures are context-independent. but nvidia's mamba2 is 67% faster at generating tokens on the exact same hardware and needs fewer flags to get there. same node, same GPU, same everything. the only variable is the model. gold medal math olympiad winner running at 187 tokens per second on single RTX 3090 a card from 6 years ago. nvidia cooked.

187,758 views • 6 months ago

bonsai 2 27b on an rtx 3060 12gb, the full receipt sheet. save this one, the 12gb row of the small gpu guide is built from it. speed by depth, then what context costs, live server, thinking on, real sessions > 7k deep: 24.4 tok/s > 12k deep: 21.9 tok/s > 35k deep: 17.8 tok/s > 77k deep: 13.0 tok/s > 64k window: 7.3gb resident > 128k window: 8.8gb resident > 192k window: 10.2gb resident > 262k window: 11.7gb resident, 0.6gb to spare, the whole native window on a 12gb card > every 64k of context costs 1.47gb, so 327k would not fit > a 41,312 token build session from 35k to 77k of context averaged 15.0 tok/s across 46 minutes > prefill 295 tok/s at 2k of context, 243 tok/s at 35k, first token in 0.6 seconds, fresh decode 26.1 tok/s > the card pinned 149 of 150 w the entire time, 78c, fan at 80%, power bound, not heat bound > 0.158 tok/s per watt at the fresh end the setup > model: ternary bonsai 2 27b, PTQ1_0, 1.75 bits per weight, 5.95gb on disk, base qwen 3.8 27b, apache 2.0 > runtime: prismml llama.cpp fork, prebuilt cuda 12.4 binary, no compile > serve: full 262k native window resident, q4 kv cache, flash attention, one slot, 11.7 of 12gb in use for anyone who followed bonsai 1 in july, that was the 3.9gb 1bit file at 42 tok/s on a 3060 ti, a faster card and a smaller file, so the same card comparison is not on the table yet, it comes with the 8gb test. what changed is the base, qwen 3.8 instead of 3.6, and the retention, 98.2% on their suite instead of 95%, and the whole 262k window fitting on 12gb.

bonsai 2 27b on an rtx 3060 12gb, the full receipt sheet. save this one, the 12gb row of the small gpu guide is built from it. speed by depth, then what context costs, live server, thinking on, real sessions > 7k deep: 24.4 tok/s > 12k deep: 21.9 tok/s > 35k deep: 17.8 tok/s > 77k deep: 13.0 tok/s > 64k window: 7.3gb resident > 128k window: 8.8gb resident > 192k window: 10.2gb resident > 262k window: 11.7gb resident, 0.6gb to spare, the whole native window on a 12gb card > every 64k of context costs 1.47gb, so 327k would not fit > a 41,312 token build session from 35k to 77k of context averaged 15.0 tok/s across 46 minutes > prefill 295 tok/s at 2k of context, 243 tok/s at 35k, first token in 0.6 seconds, fresh decode 26.1 tok/s > the card pinned 149 of 150 w the entire time, 78c, fan at 80%, power bound, not heat bound > 0.158 tok/s per watt at the fresh end the setup > model: ternary bonsai 2 27b, PTQ1_0, 1.75 bits per weight, 5.95gb on disk, base qwen 3.8 27b, apache 2.0 > runtime: prismml llama.cpp fork, prebuilt cuda 12.4 binary, no compile > serve: full 262k native window resident, q4 kv cache, flash attention, one slot, 11.7 of 12gb in use for anyone who followed bonsai 1 in july, that was the 3.9gb 1bit file at 42 tok/s on a 3060 ti, a faster card and a smaller file, so the same card comparison is not on the table yet, it comes with the 8gb test. what changed is the base, qwen 3.8 instead of 3.6, and the retention, 98.2% on their suite instead of 95%, and the whole 262k window fitting on 12gb.

31,174 views • 29 days ago

this is what 12 gigs of VRAM built in 2026. a 9 billion parameter model running on a 5 year old RTX 3060 wrote a full space shooter from a single prompt. blank screen on first try. i came back with a bug list and the same model on the same card fixed every issue across 11 files without touching a single line myself. enemies still looked wrong so i pushed another iteration and now the game has pixel art octopi, particle effects, screen shake, projectile physics and a combo system. all running locally on a card that was designed to play fortnite. three iterations. zero cloud. zero API calls. every token generated on hardware sitting under my desk. the model reads its own code, finds what's broken, patches it, validates syntax and restarts the server. i just describe what's wrong and it handles the rest. people are paying monthly subscriptions to type into a browser tab and wait for a server farm to respond. meanwhile a GPU you can find used on ebay is running a full autonomous hermes agent framework with 31 tools, 128K context window and thinking mode generating at 29 tokens per second nonstop. the game still needs work. level upgrades don't trigger and boss fights need tuning. but the fact that i'm iterating on gameplay balance instead of debugging whether the code runs at all tells you where this is headed. every iteration the game gets better on the same hardware. same 12 gigs. same 9 billion parameters. same RTX 3060 from 5 years ago your GPU is not a gaming card anymore. it's a local AI lab that never sends your data anywhere.

this is what 12 gigs of VRAM built in 2026. a 9 billion parameter model running on a 5 year old RTX 3060 wrote a full space shooter from a single prompt. blank screen on first try. i came back with a bug list and the same model on the same card fixed every issue across 11 files without touching a single line myself. enemies still looked wrong so i pushed another iteration and now the game has pixel art octopi, particle effects, screen shake, projectile physics and a combo system. all running locally on a card that was designed to play fortnite. three iterations. zero cloud. zero API calls. every token generated on hardware sitting under my desk. the model reads its own code, finds what's broken, patches it, validates syntax and restarts the server. i just describe what's wrong and it handles the rest. people are paying monthly subscriptions to type into a browser tab and wait for a server farm to respond. meanwhile a GPU you can find used on ebay is running a full autonomous hermes agent framework with 31 tools, 128K context window and thinking mode generating at 29 tokens per second nonstop. the game still needs work. level upgrades don't trigger and boss fights need tuning. but the fact that i'm iterating on gameplay balance instead of debugging whether the code runs at all tells you where this is headed. every iteration the game gets better on the same hardware. same 12 gigs. same 9 billion parameters. same RTX 3060 from 5 years ago your GPU is not a gaming card anymore. it's a local AI lab that never sends your data anywhere.

170,848 views • 7 months ago

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.

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.

126,733 views • 5 months ago

first test results are in. qwen 3.6 27b dense just banged 10 out of 10 on a single rtx 3090 24gb tier at 40 tok/s. no quant tricks. no fused kernels. just q4_k_m straight cut on llama.cpp. i wrote a particle swarm benchmark this morning, fed it the prompt, and the model autonomously built a 500 particle boids flocking system. velocity driven hue, density based brightness, trail blend rendering, mouse attraction physics, click bursts, drag paint. then it used browser automation to test its own work, found the failing tests, iterated through the code, patched tests.js, and landed all 10 green on its own. i sat there hooked for 8 minutes playing. simple but mesmerizing. mouse trails build beautiful patterns, palette cycles with space, click sends particles flying, drag paints through the swarm. simplicity that hooks you. i'll open source this prompt and the build soon so anyone can reproduce it as their own benchmark. this is the first of 5 single file agent tests i wrote for this model. four more coming. octopus invaders flagship after as final. watch the full video below. see it autonomously build from one prompt. haven't slept well since this model dropped yesterday.

first test results are in. qwen 3.6 27b dense just banged 10 out of 10 on a single rtx 3090 24gb tier at 40 tok/s. no quant tricks. no fused kernels. just q4_k_m straight cut on llama.cpp. i wrote a particle swarm benchmark this morning, fed it the prompt, and the model autonomously built a 500 particle boids flocking system. velocity driven hue, density based brightness, trail blend rendering, mouse attraction physics, click bursts, drag paint. then it used browser automation to test its own work, found the failing tests, iterated through the code, patched tests.js, and landed all 10 green on its own. i sat there hooked for 8 minutes playing. simple but mesmerizing. mouse trails build beautiful patterns, palette cycles with space, click sends particles flying, drag paints through the swarm. simplicity that hooks you. i'll open source this prompt and the build soon so anyone can reproduce it as their own benchmark. this is the first of 5 single file agent tests i wrote for this model. four more coming. octopus invaders flagship after as final. watch the full video below. see it autonomously build from one prompt. haven't slept well since this model dropped yesterday.

133,732 views • 5 months ago

okay the fuss around hermes agent is not just air. this thing has substance. installed it on a single RTX 3090 running Qwen 3.5 27B base (Q4_K_M, 262K context, 29-35 tok/s). fully local. my machine my data. first thing i did was tell it to discover itself. find its own model weights, check its own GPU, read its own server flags, and write its own identity document. it did all of it autonomously. nvidia-smi, process grep, file writes. clean execution. the TUI is genuinely premium. dark theme, ASCII art, color coded tool calls with execution times, real time streaming. you actually enjoy watching it work. 29 tools. 80 skills (that's what it reports on boot). file ops, terminal, browser automation, code execution, cron scheduling, subagent delegation. and it has persistent memory across sessions. setup took 5 minutes. one curl install, setup wizard, point to localhost:8080/v1, done. dropping qwopus for this test btw. distilled models compress reasoning and lose precision on real coding tasks. base model only from here. more experiments coming. octopus invaders (the same game that broke qwopus) will be built using hermes agent next. comparing flow and results against claude code on the same model. if you want to run local AI agents on real hardware this one deserves a serious look.

okay the fuss around hermes agent is not just air. this thing has substance. installed it on a single RTX 3090 running Qwen 3.5 27B base (Q4_K_M, 262K context, 29-35 tok/s). fully local. my machine my data. first thing i did was tell it to discover itself. find its own model weights, check its own GPU, read its own server flags, and write its own identity document. it did all of it autonomously. nvidia-smi, process grep, file writes. clean execution. the TUI is genuinely premium. dark theme, ASCII art, color coded tool calls with execution times, real time streaming. you actually enjoy watching it work. 29 tools. 80 skills (that's what it reports on boot). file ops, terminal, browser automation, code execution, cron scheduling, subagent delegation. and it has persistent memory across sessions. setup took 5 minutes. one curl install, setup wizard, point to localhost:8080/v1, done. dropping qwopus for this test btw. distilled models compress reasoning and lose precision on real coding tasks. base model only from here. more experiments coming. octopus invaders (the same game that broke qwopus) will be built using hermes agent next. comparing flow and results against claude code on the same model. if you want to run local AI agents on real hardware this one deserves a serious look.

162,824 views • 7 months ago

testing Qwen3.5-35B-A3B latest optimized version by UnslothAI on a single RTX 3090. one detailed prompt. zero handholding. watch a 3B model scaffold an entire multifile game project autonomously. the setup: > model: Qwen3.5-35B-A3B (80B total, only 3B active per token) > quant: UD-Q4_K_XL by Unsloth (MXFP4 layers removed in latest update) > speed: 112 tok/s generation, ~130 tok/s prefill > context: 262K tokens > flags: -ngl 99 -c 262144 -np 1 --cache-type-k q8_0 --cache-type-v q8_0 > engine: llama.cpp > agent: Claude Code talk to localhost:8080 (llama.cpp now has native Anthropic API endpoint. no LiteLLM needed) q8_0 KV cache cuts VRAM usage in half vs f16 at 262K. -np 1 is default but worth noting. parallel slots multiply KV cache and at 262K that's an instant OOM. the prompt was more detailed than this but you get the idea: build a space shooter with parallax backgrounds, particle systems, procedural audio, 4 enemy types, boss fights, power-up system, and ship upgrades. 8 JavaScript modules. no libraries. game's called Octopus Invaders. gameplay footage dropping next.

testing Qwen3.5-35B-A3B latest optimized version by UnslothAI on a single RTX 3090. one detailed prompt. zero handholding. watch a 3B model scaffold an entire multifile game project autonomously. the setup: > model: Qwen3.5-35B-A3B (80B total, only 3B active per token) > quant: UD-Q4_K_XL by Unsloth (MXFP4 layers removed in latest update) > speed: 112 tok/s generation, ~130 tok/s prefill > context: 262K tokens > flags: -ngl 99 -c 262144 -np 1 --cache-type-k q8_0 --cache-type-v q8_0 > engine: llama.cpp > agent: Claude Code talk to localhost:8080 (llama.cpp now has native Anthropic API endpoint. no LiteLLM needed) q8_0 KV cache cuts VRAM usage in half vs f16 at 262K. -np 1 is default but worth noting. parallel slots multiply KV cache and at 262K that's an instant OOM. the prompt was more detailed than this but you get the idea: build a space shooter with parallax backgrounds, particle systems, procedural audio, 4 enemy types, boss fights, power-up system, and ship upgrades. 8 JavaScript modules. no libraries. game's called Octopus Invaders. gameplay footage dropping next.

167,244 views • 7 months ago

this is what a 24gb VRAM builds in 2026. one prompt. ten files. 3,483 lines of code. zero handholding. i gave Qwen3.5-35B-A3B a single detailed spec describing the full game architecture and hit enter. enemy types, particle systems, procedural audio, powerups, boss fights, ship upgrades, parallax backgrounds, everything in one message. the model planned the file structure itself, wrote every module in dependency order, wired all the imports, and served the game on port 3001. it ran on first load. when it hit a bug in collision detection it read its own error output, found the issue, fixed it, and kept building. this is pure agent loop running on local hardware. what you're looking at is pixelated octopus aliens with tentacle animations, 4 layer parallax space background with planets at different depths, a full particle system handling explosions and ink splatter and engine trails and bullet impacts, procedural audio through Web Audio API with zero sound files loaded, unleash mode with combo multiplier, boss fights every 5 levels, ship upgrades that unlock as you progress. no libraries. no frameworks. vanilla JS and Canvas. 3B active parameters. single RTX 3090. llama.cpp with q8_0 KV cache at 262K context. Claude Code pointed at localhost:8080 through the native Anthropic endpoint. no API costs. 112 tok/s. a GPU you can buy used for $800. game is called Octopus Invaders and i actually like playing it.

this is what a 24gb VRAM builds in 2026. one prompt. ten files. 3,483 lines of code. zero handholding. i gave Qwen3.5-35B-A3B a single detailed spec describing the full game architecture and hit enter. enemy types, particle systems, procedural audio, powerups, boss fights, ship upgrades, parallax backgrounds, everything in one message. the model planned the file structure itself, wrote every module in dependency order, wired all the imports, and served the game on port 3001. it ran on first load. when it hit a bug in collision detection it read its own error output, found the issue, fixed it, and kept building. this is pure agent loop running on local hardware. what you're looking at is pixelated octopus aliens with tentacle animations, 4 layer parallax space background with planets at different depths, a full particle system handling explosions and ink splatter and engine trails and bullet impacts, procedural audio through Web Audio API with zero sound files loaded, unleash mode with combo multiplier, boss fights every 5 levels, ship upgrades that unlock as you progress. no libraries. no frameworks. vanilla JS and Canvas. 3B active parameters. single RTX 3090. llama.cpp with q8_0 KV cache at 262K context. Claude Code pointed at localhost:8080 through the native Anthropic endpoint. no API costs. 112 tok/s. a GPU you can buy used for $800. game is called Octopus Invaders and i actually like playing it.

153,735 views • 7 months ago

i measured qwen 3.8 27b dense, raw on a single RTX 3090 24gb GPU, before the hype finished loading. here is the honest read: 25.4 tok/s on the exact card and test where 3.6 ran 40tok/s. slower on day one, and i'm posting it anyway, because the rest of the data says watch this space. decode holds dead flat as context fills, 25 empty, 26 at 90k deep, no 27b has ever done that on my bench. one turn in the video generated 33,614 tokens and averaged 22.9 the whole way. the full 262k window sits resident in 22.2GB. the vision tower boots and the card genuinely sees. and the server logs show llama.cpp ignoring the model's MTP tensors completely, multi-token prediction trained into the weights and not wired up yet. this is the floor, measured before the optimizers woke up. tomorrow there are builders wiring MTP, tuning the hybrid attention, shipping kernels, and every commit is free speed for the same card. i benched day one so we can all watch it climb. build video, receipts, full chart in the gallery. repro below.

i measured qwen 3.8 27b dense, raw on a single RTX 3090 24gb GPU, before the hype finished loading. here is the honest read: 25.4 tok/s on the exact card and test where 3.6 ran 40tok/s. slower on day one, and i'm posting it anyway, because the rest of the data says watch this space. decode holds dead flat as context fills, 25 empty, 26 at 90k deep, no 27b has ever done that on my bench. one turn in the video generated 33,614 tokens and averaged 22.9 the whole way. the full 262k window sits resident in 22.2GB. the vision tower boots and the card genuinely sees. and the server logs show llama.cpp ignoring the model's MTP tensors completely, multi-token prediction trained into the weights and not wired up yet. this is the floor, measured before the optimizers woke up. tomorrow there are builders wiring MTP, tuning the hybrid attention, shipping kernels, and every commit is free speed for the same card. i benched day one so we can all watch it climb. build video, receipts, full chart in the gallery. repro below.

48,648 views • 2 months ago

ok this is wild. 10 year old gtx 1080 8gb pascal card running qwen3 8b locally at 18-20 tok/s via hermes agent and it's actually doing the thing. asked it to build a wireworld cellular automata simulator with 10 tests. autonomous run, no hand holding. expected it to fail on the tool calls. that's not what happened. write_file works. browser_navigate works. terminal commands work. file ops, package installs, version probes, environment setup. agent is firing tool calls cleanly and the model is reasoning about next steps at 18-20 tok/s. on hardware that pre dates "agentic" as a word. it even hit an npm install fail because node 12 is too old. didn't crash. didn't ask me. just started bootstrapping nvm on its own to fix the environment. 10 minutes in. 40% context used. 7.5gb of 8gb vram occupied. still going. i did not think this would work on this hardware. this is the most i've been wrong this month.

ok this is wild. 10 year old gtx 1080 8gb pascal card running qwen3 8b locally at 18-20 tok/s via hermes agent and it's actually doing the thing. asked it to build a wireworld cellular automata simulator with 10 tests. autonomous run, no hand holding. expected it to fail on the tool calls. that's not what happened. write_file works. browser_navigate works. terminal commands work. file ops, package installs, version probes, environment setup. agent is firing tool calls cleanly and the model is reasoning about next steps at 18-20 tok/s. on hardware that pre dates "agentic" as a word. it even hit an npm install fail because node 12 is too old. didn't crash. didn't ask me. just started bootstrapping nvm on its own to fix the environment. 10 minutes in. 40% context used. 7.5gb of 8gb vram occupied. still going. i did not think this would work on this hardware. this is the most i've been wrong this month.

92,686 views • 4 months ago

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.

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.

122,051 views • 7 months ago

watch bonsai 3.9gb model crush tool calls. this is running in hermes agent locally, and it doesn't just fire blind, it reasons first before every single action, thinks through what it actually needs, then makes the call. zero failures the whole run, not one dropped, not one malformed, across a full multi step agent loop. for anyone new here, bonsai is a 1 bit quant of qwen 3.6 27b dense, the exact model i spent all month calling the king of the 3090. PrismML took the king, crushed it down to a single sign bit per weight, 3.9gb, small enough to sit next to a gemma 12b, and it still holds an agent loop and reasons before it acts. sub 4 bit is where models are supposed to turn to mush. someone even clocked it at tool-calling 0 out of 100. not in my runs. this is the warm-up. the real test is the build gauntlet my q4 27b nailed, elements, particle swarm, mandelbrot, and openclaw runs the same one right after. we find out if 1bit actually keeps up.

watch bonsai 3.9gb model crush tool calls. this is running in hermes agent locally, and it doesn't just fire blind, it reasons first before every single action, thinks through what it actually needs, then makes the call. zero failures the whole run, not one dropped, not one malformed, across a full multi step agent loop. for anyone new here, bonsai is a 1 bit quant of qwen 3.6 27b dense, the exact model i spent all month calling the king of the 3090. PrismML took the king, crushed it down to a single sign bit per weight, 3.9gb, small enough to sit next to a gemma 12b, and it still holds an agent loop and reasons before it acts. sub 4 bit is where models are supposed to turn to mush. someone even clocked it at tool-calling 0 out of 100. not in my runs. this is the warm-up. the real test is the build gauntlet my q4 27b nailed, elements, particle swarm, mandelbrot, and openclaw runs the same one right after. we find out if 1bit actually keeps up.

55,466 views • 3 months ago

single RTX 3090. 24 GB VRAM. Qwen3.5-35B-A3B. 4-bit quant, 113 tokens per second at full 262K context harnessing Claude Code locally with no API, no subscription, no proxy. told it what it is. 30 Mamba2 layers, 10 attention, 256 experts, 8 active per token. said "build something that shows off what you can do." it visualized its own architecture. interactive. tokens flowing through layers. 256 experts lighting up on routing. served in the browser from the same GPU running inference. single prompt. then i said level up. 3D. Three.js. separate files. flythrough camera. clickable layers. it planned first, scaffolded 6 files, hit one API bug, fixed it itself, then optimized for smooth framerate. two iterations to a working 3D neural network explorer. llama.cpp just merged a native Anthropic endpoint. Claude Code points at localhost. the whole setup is two commands. no LiteLLM. no proxy config. the open source models coming out of china right now are genuinely changing what's possible on consumer hardware. respect to the Qwen team. this is acceleration.

single RTX 3090. 24 GB VRAM. Qwen3.5-35B-A3B. 4-bit quant, 113 tokens per second at full 262K context harnessing Claude Code locally with no API, no subscription, no proxy. told it what it is. 30 Mamba2 layers, 10 attention, 256 experts, 8 active per token. said "build something that shows off what you can do." it visualized its own architecture. interactive. tokens flowing through layers. 256 experts lighting up on routing. served in the browser from the same GPU running inference. single prompt. then i said level up. 3D. Three.js. separate files. flythrough camera. clickable layers. it planned first, scaffolded 6 files, hit one API bug, fixed it itself, then optimized for smooth framerate. two iterations to a working 3D neural network explorer. llama.cpp just merged a native Anthropic endpoint. Claude Code points at localhost. the whole setup is two commands. no LiteLLM. no proxy config. the open source models coming out of china right now are genuinely changing what's possible on consumer hardware. respect to the Qwen team. this is acceleration.

110,464 views • 7 months ago