MATRIX · Qwen models

Qwen releases an eight-step turbo image model

Fri, 09 Oct 2026 15:56:33 GMT

Qwen released Qwen-Image-2.1-Turbo weights. The 7B visual-generation model handles text-to-image and editing in eight denoising steps, uses the existing QwenImage21Pipeline and supports outputs up to the 2K class. The model card provides no comparative benchmark against the base model.

Recommendation: TRIAL

Sources: Qwen-Image-2.1-Turbo

Qwen releases Qwen-Image-2.1 for unified generation and editing

Sun, 20 Sep 2026 16:02:09 GMT

Qwen-Image-2.1 uses one 7B visual generator for text-to-image generation and image editing. It can generate and edit transparent RGBA assets, accept up to ten reference images, and target local edits with circles, painted annotations or masks. Diffusers, ComfyUI, vLLM-Omni, SGLang and LightX2V added day-zero support. The weights are public, but the Qwen Research License limits them to non-commercial research and evaluation.

Recommendation: TRIAL

Sources: qwen.ai · blog

Qwen3.8-LiveTranslate adds speakers and context to real-time interpreting

Sat, 19 Sep 2026 16:00:30 GMT

Qwen3.8-LiveTranslate upgrades real-time interpreting with speaker separation, bilingual source-and-translation output and long-context disambiguation. Qwen reports average lag falling from 2.8 to 2.3 seconds while retaining 60 input languages; the API produces speech in 29 languages through a 53,248-token context window. The production model is available over Alibaba Cloud Model Studio's WebSocket Realtime API in Beijing and Singapore.

Recommendation: TRIAL

Sources: qwen.ai · blog

Qwen team documents the Qwen3.8-Flash-Next architecture and its 1/9 training-FLOP economics

Wed, 02 Sep 2026 00:01:14 GMT

The Qwen team published the design and ablation study behind Qwen3.8-Flash-Next (arXiv 2608.30320, Tue 1 Sep digest): a sparse MoE with 125B parameters, 6B activated, plus 51B of n-gram embedding tables held off the accelerator. On fourteen pre-training benchmarks it beats the 397B-A17B predecessor on eight and trails on the rest by at most 2.6 points — at one-third the activated parameters, one-third the training tokens, and roughly one-ninth of the training FLOPs. Token mixing layers alternate Gated DeltaNet with global attention (one full-attention layer in four); at continued pre-training those full-attention layers are replaced by Qwen Sparse Attention, which scores context at micro-block granularity via a compressed lightweight indexer.

Recommendation: TRIAL

Sources: arXiv 2608.30320

Qwen open-weights Flash-Next, an architecture preview for Qwen4

Sat, 29 Aug 2026 11:35:46 GMT

Qwen3.8-Flash-Next is an experimental open-weight multimodal model previewing the architecture planned for Qwen4: a 125B-parameter LM with 6B activated, plus a 51B n-gram embedding table and a 4B multi-token-prediction head (~180B total in BF16). Attention is hybrid — Gated DeltaNet linear attention with Qwen Sparse Attention every 4th layer — with 262,144 tokens of native context extensible to 1M via YaRN. The card benchmarks it ahead of Qwen3.8-27B and DeepSeek-V4-Flash-0731 on agentic suites (DeepSWE 1.1 58.7, SWE-bench Pro 62.5, LiveCodeBench v6 91.9, GPQA Diamond 91.7). It ships as post-trained weights for Transformers, vLLM and SGLang under the custom qwen-community-1.0 license, with the official Qwen3.8-Flash positioned as its production counterpart.

Recommendation: WATCH

Sources: huggingface.co · Qwen3.8 Flash Next

Alibaba open-sources Qwen3.8-27B with native 262K context

Sat, 15 Aug 2026 12:26:05 GMT

Alibaba released the open weights of Qwen3.8-27B, the open sibling of the Qwen3.8 family promised at the Aug 3 Qwen3.8-Max launch. It is a dense 27.8B vision-language model with hybrid attention (Gated DeltaNet + Gated Attention), taking image, video and text input, under Apache 2.0. Native context is 262,144 tokens, extensible to 1M via YaRN. Thinking is on by default, with reasoning_effort control (xhigh/medium/low).

Recommendation: TRIAL

Sources: huggingface.co · Qwen3.8 27B