Multimodal 560B MoE (25B active) accepting text, image and video, with spatial/3D understanding and agentic abilities. Currently available via Ant's developer platform; weights not yet released.
Ant Group's multimodal mixture-of-experts model (560B total, 25B active) in the Ling family, served via API.
Updated EU status checked Last shown
Ling 3.2 Flash at a glance
- What is Ling 3.2 Flash?
- Ling 3.2 Flash is a model from Ant Group. Ant Group's multimodal mixture-of-experts model (560B total, 25B active) in the Ling family, served via API.
- Who makes Ling 3.2 Flash?
- Ling 3.2 Flash is made by Ant Group.
- Does Ling 3.2 Flash work in the EU?
- Unverified. No explicit evidence found. Last verified 14 Nov 2026.
- How much does Ling 3.2 Flash cost?
- API pricing. Available through Ant's developer platform; pricing details not confirmed.
- Where was Ling 3.2 Flash shown?
- In 1 creator video by AI Search. Most recently by AI Search on 8 Nov 2026, at 39:50 in “Weight space programming, new 3D models, Gemini Omni 1.2, GPT Image 2.6: AI NEWS”.
Sources: the vendor's own pages (linked above) and creator videos at the linked timestamps. EU status is information, not legal advice; see how we check.
Ling 3.2 Flash is a model from Ant Group's Ling family. According to a creator summary, it is a multimodal mixture-of-experts model with about 560B total parameters and 25B active per token. It accepts text, image and video input, and is described as having spatial/3D understanding and agentic abilities.
It is offered through Ant's developer platform, where the Ling model documentation lists it. Model weights had not been released as of the source material, so it can't currently be self-hosted. Official benchmarks, context length and pricing were not available in the material provided, so they are not stated here.
- Mixture-of-experts architecture, ~560B total / 25B active parameters (per creator report)
- Text, image and video input
- Spatial/3D understanding
- Agentic capabilities
- Available via Ant developer platform API; weights not yet released
As seen on YouTube
Watch the moment: the exact second top creators show it. Videos load only when you press play, via youtube-nocookie.com.
Similar AI tools
Closest neighbours by meaning, not just by category.
Ling-3.1-flash
Model · inclusionAI
Open-weights 560B hybrid-linear MoE reasoning model (25B active, 1M context) from inclusionAI, MIT licensed.
DeepSeek V4.1 Flash
Model · DeepSeek
Open-weights multimodal DeepSeek model in the V4 family, offered via API, chat and downloadable weights.
Step 3.7 Flash
Model · StepFun
Open-weights multimodal flash model from StepFun for agentic tasks involving images, interfaces, web search and long runs.
DeepSeek V4 Flash Vision Experimental
Model · DeepSeek
Experimental vision-enabled version of DeepSeek V4 Flash with image understanding; open weights under MIT, now retired on the API.
Qwen 3.8 Flash Next
Model · Alibaba
Alibaba's Qwen3.8 Flash Next: a 125B-parameter MoE language model with about 6B active parameters, previewing Qwen 4 design.
Gemini 3.8 Flash
Model · Google
Google's fast, low-cost Gemini Flash model with strong agentic coding results, plus a Cyber variant for defensive security.