This is a hybrid int4/fp8 quantized version of the Qwen3.5-122B-A10B model with full support for image and video inputs. It features sophisticated content rendering for multimodal messages and vision counting. The model balances performance and efficiency for large-scale multimodal applications.
In the Multimodal & vision space, Qwen3.5 122B A10B Int4 Fp8 Hybrid takes a focused approach. It focuses on enabling efficient multimodal inference for a very large Qwen model using hybrid quantization. It is built as an open-source project for developers. Qwen3.5 122B A10B Int4 Fp8 Hybrid is open source under the MIT license. It ships for the web and API.
bleysg builds and maintains Qwen3.5 122B A10B Int4 Fp8 Hybrid, and it first shipped in 2026. The project is developed in the open on GitHub with 38 stars and 24 commits in the last 90 days. It operates in a well-populated space: PulseGate tracks 16 similar projects. Among its 4 catalogued features are Vision Support, Video Support, and Hybrid Quantization.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do