Qwen3.8-27B-FP8 multimodal model released with Gated-DeltaNet architecture

By PulseAugur Editorial · [1 sources] · 2026-08-14 15:40

A new multimodal model named Qwen3.8-27B-FP8 has been released. This model is capable of image-text-to-text generation and features a hybrid Gated-DeltaNet/Gated-Attention architecture. It includes 64 layers, 27 billion parameters, and utilizes FP8 quantization for inference. The model's reasoning effort can be controlled through low, medium, or high settings. AI

IMPACT This release introduces a new multimodal model with controllable reasoning effort, potentially advancing image-text-to-text generation capabilities.

RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]

Read on Mastodon — mastodon.social →

model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Qwen3.8-27B-FP8 multimodal model released with Gated-DeltaNet architecture

COVERAGE [1]

Mastodon — mastodon.social TIER_1 English(EN) · aisyndicate · 2026-08-14 15:40

Qwen3.8-27B-FP8: multimodales image-text-to-text-Modell mit hybrider Gated-DeltaNet/Gated-Attention-Architektur, 64 Layer, 27B Parameter, Apache-2.0-Lizenz. FP8

Qwen3.8-27B-FP8: multimodales image-text-to-text-Modell mit hybrider Gated-DeltaNet/Gated-Attention-Architektur, 64 Layer, 27B Parameter, Apache-2.0-Lizenz. FP8-Quantisierung fuer Inferenz, Thinking-Mode per reasoning_effort (low/medium/xhigh) steuerbar. https:// huggingface.co/Q…

LINKS huggingface.co/…/Qwen3.8-27B-FP8

COVERAGE [1]

Qwen3.8-27B-FP8: multimodales image-text-to-text-Modell mit hybrider Gated-DeltaNet/Gated-Attention-Architektur, 64 Layer, 27B Parameter, Apache-2.0-Lizenz. FP8

RELATED ENTITIES

RELATED TOPICS