This is a full breakdown of Splash, the new open source local inference engine from inco_ai , built specifically around two models: Qwen 3.8 27B and…
Starting to get it dialed in a bit better, it can handle very complex instructions. Having the ability to run this locally is freaking amazing!
Check out the 2x2 panel of a level 1 Priest! Built in Three.js with Hermes Agent.
Qwen 3.8 27B on cerebras reads the battlefield, OpenCV reads cards + elixir, and Jev picks the card + placement. fresh account, starter deck.
Build and automate anything with the 2.4T parameter release.
Want the SOP? DM me. 💬
Mình vừa thử cho Qwen 3.8 27B chạy trên đúng một chiếc RTX 5090, kết hợp với Row-Bot, và đưa ra một prompt duy nhất.
The video I'm attaching is having it on xhigh reasoning so a lot of prose in thinking but it's still crazy fast - I'm really impressed!
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.
It's called Qwen 3.8 Omni Flash. It sees the screen and hears every word at the same time. Give it a long video. It finds the key moments for you.
👀 I ran Qwen 3.8 27B locally inside Hermes Desktop and gave it two real jobs: 📄 Process messy bank statement PDFs 🛠️ Call tools using Hyperframes…
Not a transcript. Not screenshots. The actual footage — sound and picture, together.
JevとGroqを組み合わせたら爆速になるのでは？と試したのですが、結果としてGroq単体で十分かつ、コストが20倍に跳ね上がる結果に...汗
✅ More than doubled the tokens/s that the M5 Pro was getting on oMLX. ✅ Two command setup ✅ Equal Game/Code quality
Grok rank Qwen 3.8 vs common cloud models.
Changes are almost instant; working on Linux, only wait due to publishing and refreshing embedded report. The video is not sped up, at all.
Qwen 3.8 Omni Flash can process: Video + audio + text + images. It watches and listens together. One model. Less tool hopping. Want the SOP? DM me.
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…
Alibaba's Qwen team has released Qwen3.8-Omni-Flash, and the ambition goes beyond another multimodal model.
Nobody is covering it. If you have two Sparks and you are not checking out Qwen3.8-Flash, you are missing out.