FRAMEWIREIndonesiaUpdated Aug 29Live wire
0:00 / 0:00

This is an example of a work created by a local Creative Agent on Qwen-3.8-Next-Flash, starting from a single picture.

The referenced picture is included in the first frame, and it has a loop configuration that returns to it at the end. I purposely do not have any say in the production, including the development.

Nobu-Kobayashi : Generative AI TechnologyAug 298
0:00 / 0:00

DeepSeek V4 Flash • Qwen 3.8 Flash

GLM 5.3 Flash • Gemini 3.7 Flash

Fabiano FirmoAug 292
0:00 / 0:00

262K context. On a 16GB RTX 5070 Ti.

🤯 Qwen 3.8 27B Q3 hits ~25 tok/s while an adaptive llama.cpp fork streams KV cache between RAM ↔ VRAM. Stock llama.cpp starts thrashing around ~120K context. Same consumer GPU. 2x+ the usable context. This could be huge for local LLMs.

SachinAug 292
0:00 / 0:00

The underlying Qwen 3.8 Max model is huge

2.4 trillion total parameters. But only around 95 billion are activated for an individual task. That mixture-of-experts approach lets it pull in the relevant parts of the model instead of activating everything.

Julian Goldie SEOAug 291