jevVIDEO
SGLang 加 /v1/systemone 让 Jev 类模型接入 TypeSafe SDK
SGLang 把 Qwen3.8-27B 变成多模态决策模型，在宝可梦火红中击败四天王与冠军，决策延迟低于 100ms，并新增 /v1/decisions 与 /v1/systemone 端点。
SGLang
@sgl_project2026.09.30收录原推发布于 2026.09.30
SGLang @sgl_project原推发布于 2026.09.30
We turned Qwen3.8-27B into a multimodal decision model. It beat Pokémon FireRed’s elite four and champion with sub-100 ms decisions from live game state. With SGLang’s native /v1/decisions, you can now turn LLMs and VLMs into classification and scoring models. We also added /v1/systemone so Jev-like open models can work with the TypeSafe SDK.
x.com/sgl_project/status/2105095379272515603采集于 2026.09.30 10:05