Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Qwopus 3.6-27b Coder I had a lot of requests to test it, so I did. I ran the same tests I’ve done on other models. It scored lower on tool-eval-bench than both Qwen 3.6 27b and Qwen 3.6 35b. It also produced broken single-file Tetris and Solar System HTML...

MiaAI_lab's profile picture

Mia

20,282 subscribers

11,183 Aufrufe • vor 2 Monaten •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

Orinth-1.0-35b vs Qwen-3.6-27b I've asked both to build a single html file of Tetris game. First here's the Orinth 35b I think it did a fine job. Every control works, the game looks fine visually. How did Qwen 3.6 27b did? See post #2
0:40

Orinth-1.0-35b vs Qwen-3.6-27b I've asked both to build a single html file of Tetris game. First here's the Orinth 35b I think it did a fine job. Every control works, the game looks fine visually. How did Qwen 3.6 27b did? See post #2

Mia

62,445 Aufrufe • vor 2 Monaten

I tested Gemma 4 12b on my Macbook Pro M5 Max with 128GB of unified memory. It runs VERY slow for a 12b parameter model. 44 tokens/second. The lava lamp that it produced was also terrible. I would stick to Qwen 3.6 27b or Qwen 3.6 35b. Google is still way behind, even with their local models.
1:36

I tested Gemma 4 12b on my Macbook Pro M5 Max with 128GB of unified memory. It runs VERY slow for a 12b parameter model. 44 tokens/second. The lava lamp that it produced was also terrible. I would stick to Qwen 3.6 27b or Qwen 3.6 35b. Google is still way behind, even with their local models.

BridgeMind

70,626 Aufrufe • vor 2 Monaten

Every big MoE model I test on my 3090 is basically to see how good it is compared to Qwen3.6-27B. "How good is it compared to Qwen3.6-27B?" is the question I get asked the most, and that shows how much people love and, more importantly, trust the 27B 27B is battle tested. It has proven its worth and has nothing left to prove. Now it just needs to make way for the next 27B. Here’s an example video: a Three.js task, same prompt, not oneshot but two shots for all models. All ran on my 3090 + 64GB VRAM Qwen-3.6-27b fully on vram while others have experts offloaded to system ram. Pinned on the left: Qwen3.6-27B-Q5 On the right, i'm switching between Inkling-Small-Q4_K_M DSV4-Flash-0731-UD-IQ1 Ling-3.0-Flash-Q4_K_M Qwen3.6-27B nailed it. The effects and visuals are just spot on.The other models got some things right, but not everything. Second place would have to go to Inkling-Small while DSV4-Flash-0731 and Ling-3.0-Flash failed to capture the full requirements. This is also why we’re all excited for Qwen Qwen3.8-27B!
2:20

Every big MoE model I test on my 3090 is basically to see how good it is compared to Qwen3.6-27B. "How good is it compared to Qwen3.6-27B?" is the question I get asked the most, and that shows how much people love and, more importantly, trust the 27B 27B is battle tested. It has proven its worth and has nothing left to prove. Now it just needs to make way for the next 27B. Here’s an example video: a Three.js task, same prompt, not oneshot but two shots for all models. All ran on my 3090 + 64GB VRAM Qwen-3.6-27b fully on vram while others have experts offloaded to system ram. Pinned on the left: Qwen3.6-27B-Q5 On the right, i'm switching between Inkling-Small-Q4_K_M DSV4-Flash-0731-UD-IQ1 Ling-3.0-Flash-Q4_K_M Qwen3.6-27B nailed it. The effects and visuals are just spot on.The other models got some things right, but not everything. Second place would have to go to Inkling-Small while DSV4-Flash-0731 and Ling-3.0-Flash failed to capture the full requirements. This is also why we’re all excited for Qwen Qwen3.8-27B!

AJ

11,670 Aufrufe • vor 23 Tagen

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

126,465 Aufrufe • vor 3 Monaten

NVIDIA CEO JENSEN HUANG UNVEILED THE RTX 3090 AND THE USED VERSION ON EBAY NOW KILLS A $400/MONTH CLAUDE CODE AND OPENAI BILL the rtx 3090 has 24gb of vram, the same as a new rtx 4090 that costs $2,000. for local ai inference vram matters more than anything else and the 3090 sells for $650-800 used it runs qwen 3.6 27b comfortably, deepseek r1 14b at 60+ tokens per second and mistral 7b for fast automation. 16gb of qwen 3.6 fits with 8gb of headroom for long context alibaba's qwen 3.6 27b beats claude 4.5 opus on realworldqa vision 84.1 vs 77.0 and on ifbench instructions 76.5 vs 58.0. a free open weight model out-benchmarks anthropic's flagship a heavy ai user pays $200 for claude code max, $200 for chatgpt pro, $20 for cursor and $20 for gemini. that's $5,280 a year and a used 3090 pays itself off in 3.5 months install ollama, pull qwen 3.6 27b, point claude code at localhost. same interface, nothing leaves the machine, nothing costs per request bookmark this and read the article below
0:33

NVIDIA CEO JENSEN HUANG UNVEILED THE RTX 3090 AND THE USED VERSION ON EBAY NOW KILLS A $400/MONTH CLAUDE CODE AND OPENAI BILL the rtx 3090 has 24gb of vram, the same as a new rtx 4090 that costs $2,000. for local ai inference vram matters more than anything else and the 3090 sells for $650-800 used it runs qwen 3.6 27b comfortably, deepseek r1 14b at 60+ tokens per second and mistral 7b for fast automation. 16gb of qwen 3.6 fits with 8gb of headroom for long context alibaba's qwen 3.6 27b beats claude 4.5 opus on realworldqa vision 84.1 vs 77.0 and on ifbench instructions 76.5 vs 58.0. a free open weight model out-benchmarks anthropic's flagship a heavy ai user pays $200 for claude code max, $200 for chatgpt pro, $20 for cursor and $20 for gemini. that's $5,280 a year and a used 3090 pays itself off in 3.5 months install ollama, pull qwen 3.6 27b, point claude code at localhost. same interface, nothing leaves the machine, nothing costs per request bookmark this and read the article below

starmex

86,410 Aufrufe • vor 2 Monaten

THIS ENGINEER RUNS CLAUDE CODE WITHOUT PAYING ANTHROPIC A SINGLE DOLLAR RTX 5090 under the desk, Qwen 3.5 35B, 140 tokens per second Claude Code pointed at localhost instead of Anthropic’s servers override two environment variables and your API bill drops to zero he built a full-stack Next.js app from scratch to prove it works then showed every bug and limitation other YouTubers hide same setup I run with two GPUs and Qwen 3.6 27B full breakdown ↓
16:33

THIS ENGINEER RUNS CLAUDE CODE WITHOUT PAYING ANTHROPIC A SINGLE DOLLAR RTX 5090 under the desk, Qwen 3.5 35B, 140 tokens per second Claude Code pointed at localhost instead of Anthropic’s servers override two environment variables and your API bill drops to zero he built a full-stack Next.js app from scratch to prove it works then showed every bug and limitation other YouTubers hide same setup I run with two GPUs and Qwen 3.6 27B full breakdown ↓

leopardracer

29,756 Aufrufe • vor 3 Monaten

Alright guys, wow. Qwen 3.8 27B is basically Qwen 4. What we have here is one of the best one-shot shark survival games I've made with any model. I ran this same prompt through fable, and the biggest thing all models struggle with in this test is the top-down view of the sharks. Fable even had all of the fins reversed and details jumbled on the first turn. Before now, it was really a 2 to 8 turn test to get something final regardless of model. Qwen 3.8 27B just nailed this in one output, no harness at all, just purely the model thoroughly thinking before outputting the final result. On every asset used, different shark variants, everything, it all looks incredibly tight. The boat movement, shark speed and engagement, everything is so much better than I expected, and I was expecting a lot. Imagine what 8 turns would do in a harness? This confirms to me that while Qwen 3.8 27B thinks a lot, for the first time in a local model on the base weights, IT MIGHT ACTUALLY BE WORTH IT This is a Frontier Lab model result, and it ran on my 5090 at 60tps in Q5_K_M. I am just in complete astoundment at what Qwen has accomplished here, and I am incredibly grateful that this model is open source. The speed will also increase drastically with optimisations. This is just an incredible day. What a blessing. More to come!
4:28

Alright guys, wow. Qwen 3.8 27B is basically Qwen 4. What we have here is one of the best one-shot shark survival games I've made with any model. I ran this same prompt through fable, and the biggest thing all models struggle with in this test is the top-down view of the sharks. Fable even had all of the fins reversed and details jumbled on the first turn. Before now, it was really a 2 to 8 turn test to get something final regardless of model. Qwen 3.8 27B just nailed this in one output, no harness at all, just purely the model thoroughly thinking before outputting the final result. On every asset used, different shark variants, everything, it all looks incredibly tight. The boat movement, shark speed and engagement, everything is so much better than I expected, and I was expecting a lot. Imagine what 8 turns would do in a harness? This confirms to me that while Qwen 3.8 27B thinks a lot, for the first time in a local model on the base weights, IT MIGHT ACTUALLY BE WORTH IT This is a Frontier Lab model result, and it ran on my 5090 at 60tps in Q5_K_M. I am just in complete astoundment at what Qwen has accomplished here, and I am incredibly grateful that this model is open source. The speed will also increase drastically with optimisations. This is just an incredible day. What a blessing. More to come!

Kyle Hessling

24,917 Aufrufe • vor 17 Tagen

You can now run Qwen 3.6 27B on <10GB of memory and retain 90% of the intelligence. Today, PrismML dropped Bonsai 27B - which takes their unique compression approach they have already successfully applied to Qwen3.6 8B and Flux2-Klein and applied it to a properly smart model. Here is it using an entire computer for deep research. Long horizon agentic calls already are strong with Qwen3.6 27B and they still are here - just on 10x less memory.
0:24

You can now run Qwen 3.6 27B on <10GB of memory and retain 90% of the intelligence. Today, PrismML dropped Bonsai 27B - which takes their unique compression approach they have already successfully applied to Qwen3.6 8B and Flux2-Klein and applied it to a properly smart model. Here is it using an entire computer for deep research. Long horizon agentic calls already are strong with Qwen3.6 27B and they still are here - just on 10x less memory.

Tim Carambat

62,879 Aufrufe • vor 1 Monat

Here's a comparison i never thought i would make Qwen-3.8-27b vs Gemini-3.7-flash Yes, you read it right! I don't know what to say, this is soo damn impressive. the difference is hugeee🤯 Take a look at this result 😍 (watch it in highest quality) Same prompt, both one shot on web-chat ui Qwen-3.8-27b produces consistent beautiful, visually stunning results. This is one of the best water surface simulations i have seen across every local models i have run and at the same time beating Gemini-3.7-flash result. And this is running on my 3090. And yes its Q5, not even f8.
2:59

Here's a comparison i never thought i would make Qwen-3.8-27b vs Gemini-3.7-flash Yes, you read it right! I don't know what to say, this is soo damn impressive. the difference is hugeee🤯 Take a look at this result 😍 (watch it in highest quality) Same prompt, both one shot on web-chat ui Qwen-3.8-27b produces consistent beautiful, visually stunning results. This is one of the best water surface simulations i have seen across every local models i have run and at the same time beating Gemini-3.7-flash result. And this is running on my 3090. And yes its Q5, not even f8.

AJ

273,466 Aufrufe • vor 17 Tagen

Parameter scaling just broke in front of me. I gave 90 math problems as images to 10 local vision models, 3 runs each, only consistent answers across all 3 runs counted. Two findings: > Gemma 4 was the most consistent family, 31B took the crown at 89.6%. > But Qwen 3.5 4B came 2 answers behind it. A 4B. Ranked 2nd out of 10. Vision-math isn't one skill, it's two: reading the image, then solving it. The real lesson for anyone running local: small ≠ weak. If you're building agentic workflows, finding the right model for each task matters more than reaching for the biggest one. In this test, the 4B ran far faster because of its size, scored higher, and left VRAM free for the rest of your stack. Full results: 🥇 Gemma 4 31B — 242/270 (89.6%) 🥈 Qwen 3.5 4B — 240/270 (88.9%) 🥉 Gemma 4 E4B — 222/270 (82.2%) 🥉 Qwen 3.6 27B — 222/270 (82.2%) 5. Gemma 4 26B A4B — 216/270 (80.0%) 6. Qwen 3.5 2B — 201/270 (74.4%) 7. Gemma 4 E2B — 192/270 (71.1%) 8. Qwen 3.6 35B A3B — 192/270 (71.1%) 9. Qwen 3.5 9B — 168/270 (62.2%) 10. Qwen 3.5 0.8B — 45/270 (16.7%) All GGUF + mmproj, Unsloth's Q6_K_XL quant.
0:44

Parameter scaling just broke in front of me. I gave 90 math problems as images to 10 local vision models, 3 runs each, only consistent answers across all 3 runs counted. Two findings: > Gemma 4 was the most consistent family, 31B took the crown at 89.6%. > But Qwen 3.5 4B came 2 answers behind it. A 4B. Ranked 2nd out of 10. Vision-math isn't one skill, it's two: reading the image, then solving it. The real lesson for anyone running local: small ≠ weak. If you're building agentic workflows, finding the right model for each task matters more than reaching for the biggest one. In this test, the 4B ran far faster because of its size, scored higher, and left VRAM free for the rest of your stack. Full results: 🥇 Gemma 4 31B — 242/270 (89.6%) 🥈 Qwen 3.5 4B — 240/270 (88.9%) 🥉 Gemma 4 E4B — 222/270 (82.2%) 🥉 Qwen 3.6 27B — 222/270 (82.2%) 5. Gemma 4 26B A4B — 216/270 (80.0%) 6. Qwen 3.5 2B — 201/270 (74.4%) 7. Gemma 4 E2B — 192/270 (71.1%) 8. Qwen 3.6 35B A3B — 192/270 (71.1%) 9. Qwen 3.5 9B — 168/270 (62.2%) 10. Qwen 3.5 0.8B — 45/270 (16.7%) All GGUF + mmproj, Unsloth's Q6_K_XL quant.

stevibe

22,882 Aufrufe • vor 3 Monaten

0xSero helped me setup local models properly and I uh, had no idea these things had gotten this good Are they frontier level? No, but considering this is running on just my 5090 it's remarkably capable First tests on a couple of programming tasks and the qwen 3.6-27b model with no reasoning feels about on par with something like sonnet 4-ish, probably better it's really impressive But also setting up local models isn't easy, I don't know nearly enough to talk much about it yet other than you need to know what you're doing to have a good experience. The out of the box stuff is not nearly as good as setting it up correctly
0:45

0xSero helped me setup local models properly and I uh, had no idea these things had gotten this good Are they frontier level? No, but considering this is running on just my 5090 it's remarkably capable First tests on a couple of programming tasks and the qwen 3.6-27b model with no reasoning feels about on par with something like sonnet 4-ish, probably better it's really impressive But also setting up local models isn't easy, I don't know nearly enough to talk much about it yet other than you need to know what you're doing to have a good experience. The out of the box stuff is not nearly as good as setting it up correctly

Ben Davis

14,490 Aufrufe • vor 3 Monaten

First impressions on Muse Glimmer! It's incredibly fast for a dense model, currently running an average of 208tps with a max of 274tps on a single 5090 with their DFLASH config. Comparatively, though, both using Open Code, Qwopus Coder (with thinking off) produced a much better shark survival game than the one I got from Glimmer. Meta's new dense model is currently just lacking some HTML canvas taste, but this is something that can be added via SFT as long as the model is stable and capable from a back-end programming perspective. And it seems to be, without a doubt. The big kicker here is that I ran this at extra high thinking, and it did not take long at all to run. Our current local leader, Qwen 27B 3.6, has a tendency to overthink, but with glimmer, that is not the case. Right now, my recommendation for general local programming (Apps, Games, Websites, Visual Tools) in this class is still Qwopus Coder with thinking disabled, or Qwopus Fusion with thinking enabled. Of course Shark Survival is a very basic domain-specific test, but I find that the result scales very well across many domains. If we're going to be shipping apps generated entirely locally, visual taste is somewhat of a bare minimum requirement, solely in my opinion, and Qwen's models in this class offer significantly more at the moment. That's actually why I initially started getting into finetuning with Qwen 3.5, they were the first base that was able to do really good front-end with some opus-trace fine-tuning. Qwen 3.6 has taste even in the base model, and we know Qwen 3.8 is going to blow us all away! Regardless, this looks like a very tempting new base model. As a first offering from Meta in this class for a long time, I am incredibly impressed and elated to have it. We now finally have a proper Single GPU frontier race, instead of us just begging Qwen for more releases. Single GPU open frontier model race is a VERY good thing. Please keep pushing Meta
1:20

First impressions on Muse Glimmer! It's incredibly fast for a dense model, currently running an average of 208tps with a max of 274tps on a single 5090 with their DFLASH config. Comparatively, though, both using Open Code, Qwopus Coder (with thinking off) produced a much better shark survival game than the one I got from Glimmer. Meta's new dense model is currently just lacking some HTML canvas taste, but this is something that can be added via SFT as long as the model is stable and capable from a back-end programming perspective. And it seems to be, without a doubt. The big kicker here is that I ran this at extra high thinking, and it did not take long at all to run. Our current local leader, Qwen 27B 3.6, has a tendency to overthink, but with glimmer, that is not the case. Right now, my recommendation for general local programming (Apps, Games, Websites, Visual Tools) in this class is still Qwopus Coder with thinking disabled, or Qwopus Fusion with thinking enabled. Of course Shark Survival is a very basic domain-specific test, but I find that the result scales very well across many domains. If we're going to be shipping apps generated entirely locally, visual taste is somewhat of a bare minimum requirement, solely in my opinion, and Qwen's models in this class offer significantly more at the moment. That's actually why I initially started getting into finetuning with Qwen 3.5, they were the first base that was able to do really good front-end with some opus-trace fine-tuning. Qwen 3.6 has taste even in the base model, and we know Qwen 3.8 is going to blow us all away! Regardless, this looks like a very tempting new base model. As a first offering from Meta in this class for a long time, I am incredibly impressed and elated to have it. We now finally have a proper Single GPU frontier race, instead of us just begging Qwen for more releases. Single GPU open frontier model race is a VERY good thing. Please keep pushing Meta

Kyle Hessling

17,399 Aufrufe • vor 21 Tagen

HERMES AGENT CROSSED 140,000 GITHUB STARS IN 3 MONTHS AND JUST BECAME THE MOST USED AGENT IN THE WORLD. Most AI agents forget everything between sessions. Hermes writes its own skills from experience. Next time it runs the skill, improves it, and gets faster. Independent benchmarks show agents with 20+ self-created skills complete similar tasks 40% faster than fresh instances. Qwen 3.6 where the 35B version outperforms last year's 120B models at one third the memory footprint. DGX Spark with 128GB unified memory running everything locally at $0 per month after hardware. The setup takes 30 minutes. LM Studio plus Qwen 3.6 27B for the model server. One install script for Hermes. One config connecting them. Set context window to 65,536 tokens or nothing works. After one month of daily use your skills directory has 20 to 50 learned workflows. Your Hermes is genuinely different from anyone else's.
0:25

HERMES AGENT CROSSED 140,000 GITHUB STARS IN 3 MONTHS AND JUST BECAME THE MOST USED AGENT IN THE WORLD. Most AI agents forget everything between sessions. Hermes writes its own skills from experience. Next time it runs the skill, improves it, and gets faster. Independent benchmarks show agents with 20+ self-created skills complete similar tasks 40% faster than fresh instances. Qwen 3.6 where the 35B version outperforms last year's 120B models at one third the memory footprint. DGX Spark with 128GB unified memory running everything locally at $0 per month after hardware. The setup takes 30 minutes. LM Studio plus Qwen 3.6 27B for the model server. One install script for Hermes. One config connecting them. Set context window to 65,536 tokens or nothing works. After one month of daily use your skills directory has 20 to 50 learned workflows. Your Hermes is genuinely different from anyone else's.

Lummox

69,676 Aufrufe • vor 2 Monaten

Qwen 3.8 27B beat Claude Opus 4.6 for $0 open-weights Qwen 3.8 27B on one rented GPU vs anthropic's Claude Opus 4.6. same prompt, no edits the test: one self-contained three.js file rendering sunken Atlantis — temple, columns, boid fish, kelp, bubbles, god rays, caustics — plus a 20-second camera flythrough cost: Qwen 3.8 27B — free (open weights, self-hosted) Claude Opus 4.6 — $0.21 what we saw: • qwen actually built a temple — full colonnade, domed roof, glowing portal centered. opus made a long row of columns fading into fog • qwen's portal is the star of the shot. opus leaves it small and off to the side • qwen packed more into frame — fish, kelp, ruins layered together. opus spreads thin and empty qwen's weights are on hugging face — you can run this one yourself!
0:20

Qwen 3.8 27B beat Claude Opus 4.6 for $0 open-weights Qwen 3.8 27B on one rented GPU vs anthropic's Claude Opus 4.6. same prompt, no edits the test: one self-contained three.js file rendering sunken Atlantis — temple, columns, boid fish, kelp, bubbles, god rays, caustics — plus a 20-second camera flythrough cost: Qwen 3.8 27B — free (open weights, self-hosted) Claude Opus 4.6 — $0.21 what we saw: • qwen actually built a temple — full colonnade, domed roof, glowing portal centered. opus made a long row of columns fading into fog • qwen's portal is the star of the shot. opus leaves it small and off to the side • qwen packed more into frame — fish, kelp, ruins layered together. opus spreads thin and empty qwen's weights are on hugging face — you can run this one yourself!

AI/ML API

102,960 Aufrufe • vor 17 Tagen

nemotron 3 omni q8 on dgx spark 128gb vram cranking via hermes agent at 56 tok/s. first night of real local agentic on this box and local hits harder than i thought it would. q8 (near lossless quant, perplexity loss <1% vs fp16) running 256k context on 33 gb of unified memory, 90+ gb still free. multimodal omni weights included. hermes agent driving from telegram, talking to it from bed. speed: 56 tok/s generation, 1,300 tok/s prefill. for context, qwen 3.6 27b at q4 (heavy quant) on 3090 = 40 tok/s. nemotron at higher precision quant on spark beats qwen at lower precision quant on 3090. moe 3.5b active params architecture earns its keep. what i tested tonight: agentic tool calling works clean. ask it to check disks, it autonomously runs df -h through hermes agent. ask it to set up telegram gateway, it invokes the hermes-agent skill, walks through the prompts, completes the flow. overthinks a bit before tool calls (reasoning model trait) but lands the right move every time. researches api docs, internalizes, tests, documents. completes tasks. current models on dgx spark: 9 gguf files, 305 gb total, mix of qwen 3.6 27b dense (5 quants), nemotron omni (4 quants), deepseek v4-flash 158b q4 (the 112gb flagship test). more data coming this week as i benchmark each.
1:40

nemotron 3 omni q8 on dgx spark 128gb vram cranking via hermes agent at 56 tok/s. first night of real local agentic on this box and local hits harder than i thought it would. q8 (near lossless quant, perplexity loss <1% vs fp16) running 256k context on 33 gb of unified memory, 90+ gb still free. multimodal omni weights included. hermes agent driving from telegram, talking to it from bed. speed: 56 tok/s generation, 1,300 tok/s prefill. for context, qwen 3.6 27b at q4 (heavy quant) on 3090 = 40 tok/s. nemotron at higher precision quant on spark beats qwen at lower precision quant on 3090. moe 3.5b active params architecture earns its keep. what i tested tonight: agentic tool calling works clean. ask it to check disks, it autonomously runs df -h through hermes agent. ask it to set up telegram gateway, it invokes the hermes-agent skill, walks through the prompts, completes the flow. overthinks a bit before tool calls (reasoning model trait) but lands the right move every time. researches api docs, internalizes, tests, documents. completes tasks. current models on dgx spark: 9 gguf files, 305 gb total, mix of qwen 3.6 27b dense (5 quants), nemotron omni (4 quants), deepseek v4-flash 158b q4 (the 112gb flagship test). more data coming this week as i benchmark each.

Sudo su

30,228 Aufrufe • vor 4 Monaten

been playing with hermes agent paired with qwen 3.5 dense 27B on my single 3090 since last night. there is something about this harness that caught me and i think i know what it is. i've now run five qwen configs on consumer hardware: 35B MoE (3B active) -- 112 tok/s flat across 262K context, 1x 3090 27B dense -- 35 tok/s, zero degradation across the same range, 1x 3090 qwopus 27B (opus distilled) -- 35.7 tok/s, same architecture, different brain 80B coder -- 46 tok/s on 2x 3090s, oneshotted a 564 line particle sim 80B coder -- 1.3 tok/s on 1x 3090, bleeding through RAM because it didn't fit but it still ran with same benchmarks. same prompts. same quant where possible. every config is documented. i know these models. and hermes agent is the first harness that feels like it respects that work. tool calls show inline with execution time. nvidia-smi 0.2s. write_file 0.7s. you see exactly what the agent is doing and how long each step takes. no mystery. no black box. no tool call failures so far and i've been pushing it. most agent frameworks feel like you're watching a spinner and hoping. hermes shows the work. that transparency changes how you trust the output. once you use it you see the UX decisions are not accidental. Teknium 🪽 and the nous team built this like engineers who actually use their own tools. 80 skills. 29 tools. persistent memory. context compression. runs clean on a single consumer GPU.
2:12

been playing with hermes agent paired with qwen 3.5 dense 27B on my single 3090 since last night. there is something about this harness that caught me and i think i know what it is. i've now run five qwen configs on consumer hardware: 35B MoE (3B active) -- 112 tok/s flat across 262K context, 1x 3090 27B dense -- 35 tok/s, zero degradation across the same range, 1x 3090 qwopus 27B (opus distilled) -- 35.7 tok/s, same architecture, different brain 80B coder -- 46 tok/s on 2x 3090s, oneshotted a 564 line particle sim 80B coder -- 1.3 tok/s on 1x 3090, bleeding through RAM because it didn't fit but it still ran with same benchmarks. same prompts. same quant where possible. every config is documented. i know these models. and hermes agent is the first harness that feels like it respects that work. tool calls show inline with execution time. nvidia-smi 0.2s. write_file 0.7s. you see exactly what the agent is doing and how long each step takes. no mystery. no black box. no tool call failures so far and i've been pushing it. most agent frameworks feel like you're watching a spinner and hoping. hermes shows the work. that transparency changes how you trust the output. once you use it you see the UX decisions are not accidental. Teknium 🪽 and the nous team built this like engineers who actually use their own tools. 80 skills. 29 tools. persistent memory. context compression. runs clean on a single consumer GPU.

Sudo su

100,259 Aufrufe • vor 5 Monaten

Local models just got a free intelligence upgrade. Someone took the new Qwen3.8-27B (the one that already runs in 22GB) and only changed the chat template. No fine-tune. No new training. Just a smarter system prompt that kills all the filler and forces the model to stay sharp. Result: the same weights are now fixing real post-cutoff bugs faster and cleaner than Claude Opus 5 High in their tests. They call it the Sharp template. It’s free. Drop it on any Qwen3.5 / 3.6 / 3.8 model and it just works with llama.cpp. This is the kind of quiet upgrade that actually moves the needle for people running models locally.
0:39

Local models just got a free intelligence upgrade. Someone took the new Qwen3.8-27B (the one that already runs in 22GB) and only changed the chat template. No fine-tune. No new training. Just a smarter system prompt that kills all the filler and forces the model to stay sharp. Result: the same weights are now fixing real post-cutoff bugs faster and cleaner than Claude Opus 5 High in their tests. They call it the Sharp template. It’s free. Drop it on any Qwen3.5 / 3.6 / 3.8 model and it just works with llama.cpp. This is the kind of quiet upgrade that actually moves the needle for people running models locally.

Harman

73,661 Aufrufe • vor 12 Tagen

I’ve tried driving Qwen 3.6 on my MacBook Pro with a few different agent harnesses: Claude Code: too slow. It moves like a turtle. Codex: Qwen stopped in the middle of tasks, probably due to tool-call mismatches between them. Qwen Code: so far the best experience. Very smooth. Qwen Code sometimes even tells a joke while you’re waiting. This video is played at original speed so you can get a real feel for how Qwen works.
0:47

I’ve tried driving Qwen 3.6 on my MacBook Pro with a few different agent harnesses: Claude Code: too slow. It moves like a turtle. Codex: Qwen stopped in the middle of tasks, probably due to tool-call mismatches between them. Qwen Code: so far the best experience. Very smooth. Qwen Code sometimes even tells a joke while you’re waiting. This video is played at original speed so you can get a real feel for how Qwen works.

Michael Guo

20,505 Aufrufe • vor 3 Monaten

hey if you're thinking about running qwopus (the claude opus distilled qwen 3.5 27B) as a coding agent, this might save you a few hours. i tested both the base and the distilled version on the same hardware. single RTX 3090. same prompt. same context. same everything. the only variable was the model weights. base qwen 3.5 27B built octopus invaders in 13 minutes. 1,827 lines across 11 files. zero steering. one scope bug that took 2 lines to fix. game ran. qwopus couldn't finish the same task. enemies overlapping on screen. bullets not firing. controls worked but the game was broken. i had to steer it multiple times and it still didn't produce a playable result. both run at 35 tok/s. both use thinking mode. the distilled version actually has better jinja compatibility and doesn't stall midtask like base does on claude code. for conversation and reasoning it feels sharper. but for multifile autonomous coding where the model needs to coordinate 10+ files without losing track, base wins and it's not close. distillation compresses reasoning patterns but seems to lose precision on complex coordination. the model "thinks" well but can't hold the full picture across files the way base can. tested on opencode (base) and claude code (both). next up is hermes agent framework on base. same hardware. same prompt. comparing agents now, not just models. video below. first half is the distilled model's broken game. second half is what base built on the same 3090. judge for yourself.
0:20

hey if you're thinking about running qwopus (the claude opus distilled qwen 3.5 27B) as a coding agent, this might save you a few hours. i tested both the base and the distilled version on the same hardware. single RTX 3090. same prompt. same context. same everything. the only variable was the model weights. base qwen 3.5 27B built octopus invaders in 13 minutes. 1,827 lines across 11 files. zero steering. one scope bug that took 2 lines to fix. game ran. qwopus couldn't finish the same task. enemies overlapping on screen. bullets not firing. controls worked but the game was broken. i had to steer it multiple times and it still didn't produce a playable result. both run at 35 tok/s. both use thinking mode. the distilled version actually has better jinja compatibility and doesn't stall midtask like base does on claude code. for conversation and reasoning it feels sharper. but for multifile autonomous coding where the model needs to coordinate 10+ files without losing track, base wins and it's not close. distillation compresses reasoning patterns but seems to lose precision on complex coordination. the model "thinks" well but can't hold the full picture across files the way base can. tested on opencode (base) and claude code (both). next up is hermes agent framework on base. same hardware. same prompt. comparing agents now, not just models. video below. first half is the distilled model's broken game. second half is what base built on the same 3090. judge for yourself.

Sudo su

45,052 Aufrufe • vor 5 Monaten

watch anon! 27 billion parameter model dancing at 42 tokens a second on a used rtx 3060ti, that costs 200 dollars. consuming 159 watts. temp 62 degrees. pushing almost 100% on the gpu. 6.8 of 8 gigs used with a 128k context window loaded. and every one of those spikes is the model reasoning and firing tool calls. the model is bonsai 27b at 1 bit. PrismML took qwen 3.6 27b, and crushed it to 3.9 gigabytes, about a single sign bit per weight. it drops onto an 8gb rtx 3060 ti, an ampere card from 2020, and it doesn't just load, it serves, it holds the full 128k window, it runs a hermes agent for half an hour straight, it hits its own failures and recovers, it never once drops the loop. the absurd part is the power meter. that same 27b, in its normal quant 4, wants a 24gb card and nearly 17 gigs of weights. this is it at 3.9, doing agentic work at 160 watts, on the tier people call budget. the numbers are sitting right there on nvtop, not in a slide, not a claim, the meter reading while it works. so i keep landing on one question. the model is open, apache 2.0. the card is old and cheap. the whole thing runs offline. why is almost nobody actually running this. is it the hardware, the setup, or did nobody tell you a 27b fits on the card already in your closet.
0:26

watch anon! 27 billion parameter model dancing at 42 tokens a second on a used rtx 3060ti, that costs 200 dollars. consuming 159 watts. temp 62 degrees. pushing almost 100% on the gpu. 6.8 of 8 gigs used with a 128k context window loaded. and every one of those spikes is the model reasoning and firing tool calls. the model is bonsai 27b at 1 bit. PrismML took qwen 3.6 27b, and crushed it to 3.9 gigabytes, about a single sign bit per weight. it drops onto an 8gb rtx 3060 ti, an ampere card from 2020, and it doesn't just load, it serves, it holds the full 128k window, it runs a hermes agent for half an hour straight, it hits its own failures and recovers, it never once drops the loop. the absurd part is the power meter. that same 27b, in its normal quant 4, wants a 24gb card and nearly 17 gigs of weights. this is it at 3.9, doing agentic work at 160 watts, on the tier people call budget. the numbers are sitting right there on nvtop, not in a slide, not a claim, the meter reading while it works. so i keep landing on one question. the model is open, apache 2.0. the card is old and cheap. the whole thing runs offline. why is almost nobody actually running this. is it the hardware, the setup, or did nobody tell you a 27b fits on the card already in your closet.

Sudo su

23,391 Aufrufe • vor 1 Monat