Releases·Model Release
inclusionAI releases Ling-3.0-flash-VL: 124B-parameter sparse MoE vision-language model with 1M-token context
inclusionAI released Ling-3.0-flash-VL on Hugging Face, a vision-language model adding native image and video understanding to Ling-3.0-flash's language and long-context capabilities. It uses a sparse MoE design (124B total, 5.5B active per token) with a 1M-token context window, aimed at visual reasoning and agentic workflows.
183 engagement2 sourcesTue, Sep 8, 2026, 03:57 PM
AI Overview
- Ling-3.0-flash-VL extends Ling-3.0-flash with native image and video understanding while inheriting its language, reasoning, and long-context capabilities, per the Hugging Face model card.
- The model totals 124B parameters but activates only 5.5B per token via sparse MoE, and supports context windows up to 1M tokens.
- A ViT visual encoder plus two-layer MLP projector align image/video features with text for unified multimodal reasoning.
- VideoRoPE encodes spatial positions and temporal order, supporting event localization, long-video QA, and video clip editing.
- A 42-layer hybrid backbone alternates KDA and Gated MLA layers at a 5:1 ratio for efficient long-context processing across text, images, video, and agent task histories.
Related
Releases·Model Releasefeatured
Ant Group's inclusionAI open-weights Ling-3.0-flash under MIT, 124B total with 5.1B active
Ant Group's lab inclusionAI released Ling-3.0-flash, a 124B-parameter hybrid-reasoning MoE that activates only 5.1B per token, under a plain MIT license. The lab claims it matches or beats its own trillion-param flagship Ring-2.6-1T. A smaller 8B variant, Ling-3.0-tiny, followed on Aug 10.
Ant Group's lab inclusionAI released Ling-3.0-flash, a 124B-parameter hybrid-reasoning MoE that activates only 5.1B per token, under a plain MIT license. The lab claims it matches or beats its own trillion-param flagship Ring-2.6-1T. A smaller 8B variant, Ling-3.0-tiny, followed on Aug 10.
Updated Mon, Aug 10, 2026, 05:11 PM — Ling-3.0-tiny (8B A1.3B MoE) open-weighted Aug 10 with reported FP8 speeds on DGX Spark and M4 Pro.
340920 engagement7 sourcesreddit, rssTue, Aug 4, 2026, 12:00 AM
Releases·Model Release
Microsoft releases Mage-VL, a 4B codec-native streaming multimodal model
Microsoft released Mage-VL, a 4B-parameter multimodal foundation model for image and video understanding with a visual encoder trained entirely from scratch. It uses a codec-native, proactive-streaming approach to handle real-time perception efficiently, addressing the Moravec's paradox where VLMs are strong at offline reasoning but slow on simple streaming tasks.
Microsoft released Mage-VL, a 4B-parameter multimodal foundation model for image and video understanding with a visual encoder trained entirely from scratch. It uses a codec-native, proactive-streaming approach to handle real-time perception efficiently, addressing the Moravec's paradox where VLMs are strong at offline reasoning but slow on simple streaming tasks.
2 engagement1 sourceredditTue, Jul 28, 2026, 06:47 PM
Developments·Paper
Mage-VL: Codec-native streaming multimodal model targets real-time perception
Researchers introduce Mage-VL, a codec-native streaming foundation model for real-time multimodal understanding. Its custom tokenizer, Mage-ViT, selectively encodes dynamic regions using motion vectors and residual energy, addressing the inefficiency of standard VLMs in streaming tasks.
Researchers introduce Mage-VL, a codec-native streaming foundation model for real-time multimodal understanding. Its custom tokenizer, Mage-ViT, selectively encodes dynamic regions using motion vectors and residual energy, addressing the inefficiency of standard VLMs in streaming tasks.
0 engagement1 sourcearxivMon, Jul 27, 2026, 05:59 PM
Releases·Model Release
CohereLabs releases North Micro Vision Instruct, a 2.4B open-weight VLM
CohereLabs released North Micro Vision Instruct, a 2.4B-parameter open-weight vision-language model under Apache 2.0. It supports native-resolution images, multilingual and multi-image inputs, and is designed for prototyping and fine-tuning.
CohereLabs released North Micro Vision Instruct, a 2.4B-parameter open-weight vision-language model under Apache 2.0. It supports native-resolution images, multilingual and multi-image inputs, and is designed for prototyping and fine-tuning.
122 engagement1 sourceredditWed, Aug 12, 2026, 04:50 PM
Releases·Model Release
Agnes-AI releases Agnes-3.0-Flash 33B open-weights multimodal model with 262K context and hybrid attention
Agnes-AI published Agnes-3.0-Flash, a 33B open multimodal model with a 262,144-token context window, adjustable reasoning effort, tool calling, and text, image, and video understanding. It uses a hybrid architecture where only 18 of 72 layers run global attention, keeping most state recurrent. Artificial Analysis lists it publicly despite the org labeling similar models proprietary.
Agnes-AI published Agnes-3.0-Flash, a 33B open multimodal model with a 262,144-token context window, adjustable reasoning effort, tool calling, and text, image, and video understanding. It uses a hybrid architecture where only 18 of 72 layers run global attention, keeping most state recurrent. Artificial Analysis lists it publicly despite the org labeling similar models proprietary.
298 engagement1 sourceredditSat, Sep 12, 2026, 08:05 AM