This is the Pelican Test using Qwen 3.8 W4A16 Flash Next running locally.
Looks pretty good to me. What do you think?
Looks pretty good to me. What do you think?
This works in Hermes + OMP and other coding harnesses. It built this retro cube game PLUS enabled it over my tailscale for me.
Succede: ChatGPT, Gemini e Claude hanno regole interne che bloccano le richieste rischiose.
We can use Alibaba’s new 125B model for free through Qoder until September 30.
Qwen 3.8 • Muse Glimmer 30b • Qwen 3.8 Uncensored All models run in a hardware-secured enclave so nobody can see, store or train on your prompts.
Ask the chat for code → paste it into an editor → hit an error → paste the error back → repeat.
Calls, videos, meetings, images, and documents can all go into one workflow. The basic loop: → Feed in audio, video, text, and images.
Compressed model from Qwen 3.8 27B at 3.5bpw only 11.5GiB fit on single GPU RTX 3090 at full context 256k.
First tests !! CompfyUI + ChatGPT6 (make that json for ...) + MBP M3 max 128 GB Next : add sound, then voice, then lipssync, then improve quality,…
It sees the screen. It hears every word. And it never loses the plot.
No transcripts. No screenshots. No doing half the work yourself. It's called Qwen 3.8 Omni Flash.
Model: "Maaf, saya tidak bisa membantu dengan itu." 🙃 Qwen 3.8 27B Uncensored: uncensored penuh, 131K context, chat/coding/agent, cache read Rp…
Gemini wins end to end, Qwen lands at 84% of its HOTA. The biggest lever was not the model at all. Write-up:
ONE DGX Spark Working with MiaAI_lab recipe of Qwen 3.8 flash next, but you can use any model Simple test of an essay about the research paper:
Two boxes running the same Gemma 4 12B IT QAT at 4-bit, and the spread between them is 3%. 29 tok/s on Strix Halo (Q4_K_XL).
At launch a little more than a month ago the model was running around ~20tk/s It’s 6x faster now, local AI is progressing at lightning speed!
Fortunately, there’s a very simple and elegant way to reduce or perhaps even completely eliminate them. 🧵
So I built Tesrune for the Bitget_AI S2 Hackathon. Tesrune is a dark-hours AI trading desk for US equities. You add the stocks you're holding.