Ant Group 发布医疗模型 LingSante-3.1-flash,开放模型权重Chinese only
LingSante-3.1-flash 是总参数量为 5600 亿的医疗混合专家模型,每个 token 激活 250 亿参数。发布方称,它在 HealthBench Professional 等三项评测中超过 GPT-6 Astra、Claude Opus 5.5 和 Gemini 4 Argon;模型权重现已开放。
1 developments3 reports1 sources
Event developments 1
Ant Group releases medical model LingSante-3.1-flash with available weights
The mixture-of-experts model has 560 billion total parameters and activates 25 billion per token. Its publisher reports that it outperformed GPT-6 Astra, Claude Opus 5.5 and Gemini 4 Argon on HealthBench Professional, HLE and BrowseComp. The model weights are available; the benchmark results are publisher-reported.
View source reports · 1 sources · 3 reports
- Rohan Paul (@rohanpaul_ai) · LingSante-3.1-flash medical model weights are available
- Rohan Paul (@rohanpaul_ai) · Ant Group releases medical model LingSante-3.1-flash with available weights
- Rohan Paul (@rohanpaul_ai) · Ant Group releases LingSante-3.1-flash medical model and makes weights available
All reports 3
Ant Group releases LingSante-3.1-flash medical model and makes weights available
LingSante-3.1-flash is a medical mixture-of-experts model with 560 billion total parameters and 25 billion activated per token. Its publisher says it outperforms GPT-6 Astra, Claude Opus 5.5 and Gemini 4 Argon on three benchmarks, including HealthBench Professional. The model weights are available.
1 more source reported this
- Rohan Paul (@rohanpaul_ai)
Ant Group releases medical model LingSante-3.1-flash with available weights
The mixture-of-experts model has 560 billion total parameters and activates 25 billion per token. Its publisher reports that it outperformed GPT-6 Astra, Claude Opus 5.5 and Gemini 4 Argon on HealthBench Professional, HLE and BrowseComp. The model weights are available; the benchmark results are publisher-reported.
1 more source reported this
- Rohan Paul (@rohanpaul_ai)
LingSante-3.1-flash medical model weights are available
LingSante-3.1-flash is a medical mixture-of-experts model with 560 billion total parameters and 25 billion activated per token, and its weights are available. Its publisher says it combines complex medical reasoning and long-horizon workflows with general capabilities, and claims leading results on benchmarks including HealthBench Professional.
1 more source reported this
- Rohan Paul (@rohanpaul_ai)