Reportedly: Ant Group's Ling-3.1-flash launches with 560B total parameters and 1M context
Reported by IT Home · not yet confirmed by the company or a second independent outlet. We update this page when it is.
Ant Group's Ling-3.1-flash model launched with about 560B total parameters and 25B active parameters per token, IT Home reported. The model has a 1M context window, but free trial service is limited to 256K for two weeks. After the free period, the team plans to open source the model and enable 1M context.
China context
- Original name
- 蚂蚁百灵
- Outside China
- Not stated in the sources yet
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Developers can test the model during the free trial, but outside-China API access is not confirmed; check the official Ant Group channels for availability.
- For investors
- If the model is open-sourced as planned, it may intensify competition in the open-weights market, affecting pricing and adoption of long-context models.
Translated from Chinese. Quotes and facts link to the original sources.
Ant Group's Ling-3.1-flash model was released with approximately 560B total parameters and 25B active parameters per token, according to IT Home. The model supports a maximum context window of 1M, but during a two-week free trial period, the service length is 256K. After the free trial ends and the service becomes paid, the team plans to open source the model and open the 1M context.
The model uses a mixture-of-experts architecture with about 560B total parameters and 25B active parameters per token, enabling efficient inference. The context window is 1M, but the free trial is limited to 256K. The company reports optimization for general agents, search, office work, software development, and improvements in medical, financial, and materials science scenarios.
Developers outside China may gain access to a large open-weights model with 1M context after the free trial, potentially reducing costs for long-context applications. Competitors offering long-context models will face a new open-source option if the planned open-sourcing occurs.
The free trial offers developers a chance to evaluate the model's capabilities on long-context tasks at no cost for two weeks. If open-sourced, it could lower the barrier for enterprises needing large context windows without licensing fees.
The specific signal to check is whether Ant Group releases the model weights on a public platform like Hugging Face after the free trial period. Also verify if the 1M context becomes available to all users or only paid tiers.