On my Strix Halo PC, Qwen 3.8 Flash-Next on Halogen passed 66/72 coding jobs and generated at 38.0 tok/s versus 13.9 in my earlier setup.
Different quant, backend and drafting make this a configuration comparison, not an engine-only test. Runs:
Qwen 3.8 flash next built this 3d octopus invaders game in about 2 hours on my 2x dgx sparks and i didn't even try hard.
The model is soo tasteful you can feel it in responses. here are the data and specs: qwen 3.8 flash next, 180b moe, official fp8 45 tok/s fresh, 35…
Qwen 3.8 Flash-Next resolved 17/18 coding jobs on my Strix Halo PC.
But as context grew, generation fell from about 15 to 8 tok/s across three harnesses.
New Mac Studio day one: Qwen 3.8 27B, fully local, running on Bionic.
The speed is absolutely CRAZY 🤯 Local AI is no longer a compromise.
Kan een lokaal AI model 3d modellen maken in three.js?
Ik teste het met Qwen 3.8 27b op mijn laptop! De opdracht: maak een robot met uitwisselbare onderdelen, 4 animaties en exploded view.
Opensource Diary - Day 2.
Made with Tesseract by Mirage Model - Qwen 3.8 27b Hardware - Single Nvidia 96 RTX 6000 Workstation Edn Rate my edits... today...yikes
Creating a simple tool to easily download files from the Internet Archive with Qoder (Qwen 3.8 Flash).
You can access it for free and as an open-source project at qoder_ai_ide
You can test some of the biggest AI models for FREE 😳
1M free tokens to play with models like: - Claude Opus 5.5 - Claude Fable 5.1 - GPT-5.6 Sol - GPT-5.6 Luna - Grok 4.6 - Qwen 3.8 Max - DeepSeek V4…
Ik liet Qwen 3.8 27b een chique website bouwen voor Studio Fleur, een kwaliteitsbloemist in Antwerpen.
Het model draaide lokaal, bouwde zelf de website en genereerde de afbeeldingen via api. Ongeveer een half uurtje werk, met resultaat!
Pasar malam paling rame minggu ini?
Pasar Model Vikey Top 5 model paling laris di Vikey: 🥇 Qwen 3.8 27B Uncensored: jawabnya lugas, no ceramah 🥈 DeepSeek V4 Flash: input cuma Rp1.080…
Trust no one · not even noone
⊘⬡ 🜍 🜂 A Leaderboard Is Not the Country Full Gate-Empty lens on where China’s AI sits · holes named, not averaged TN-CHINA-AI-WHERE-π-001 rev.b…
Bouw nu zelf je POS kassa systeem, lokaal met opensource ai!
Geef Qwen 3.8 27b een lijst met je producten, barcodes en prijzen, en binnen een half uur kun je scannen en facturen sturen!
V2 updated: 348k voxels - 40% bigger island beaches, more details all around, time of day additions.
EXL3 - Qwen 3.8 Flash next - single spark. <100k tokens decode total.
Ok here's my voxel pagoda "Thai Island Paradise" - Qwen 3.8 Flash Next using the EXL3 engine (credit…
Ok here's my voxel pagoda "Thai Island Paradise" - Qwen 3.8 Flash Next using the EXL3 engine (credit ViC305 ) which I patched for Hermes…
This is so freaking sick!
I love botting. i’m going through and improving the gym and world model. qwen 3.8 27b on my rtx5090 with jev for fighting.
Looking forward to it!
Current OpenClaw doing a 7sec run with Sol as primary, Luna as light and Jev (or Qwen 3.8 Flash Next just 2sec slower) searching for a flight in a…
I just ran the largest speed and capacity test for almost every entry AI subscription.
How did Opus 5.5, GPT 6 Sol and Grok 4.7 etc. stack up? Let's find out!
Z.ai Deploys Accelerated GLM 5.3 Prime
GLM 5.3 Prime is the high-speed variant of GLM-5.3, live today on OpenRouter.
Alibaba is giving away a 125B parameter flagship model for free.
No catch found yet. Qwen 3.8-Flash. 1M token context, enough to feed it an entire codebase or a full-length book in one shot.
One shot builds aren’t cloud only.
Qwen 3.8 Flash plans the build, then hands it to Qwen 3.8 27B, all running locally across the GPU, CPU and 128GB of unified memory on one AMD Ryzen…