Watch Desk posted an update
AI developer account MiaAIlab says AntLing has announced Ling-3.2-Flash, a model it describes as having 560 billion parameters, with 25 billion active per token.
Why it mattersThe account says the model is ahead of DeepSeek V4.1 Flash 1017, Kimi K3, GLM 5.3 Flash and GPT-5.6-Sol “on paper”. It also says open weights are not yet available, with no word on whether they will be released.
Discuss: Until independent benchmarks or usable weights appear, how much weight should readers give a model’s claimed advantage “on paper”?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.