Liquid AI ships d1, a decision model for structured software tasks
- Source
- x.com
- Date
Announcing d1, our first decision model. It's the first model to outperform Jev on @huggingface's Decision Index. > wins on multilingual evals > more robust against prompt injection > handles longer inputs more effectively > built for fast, structured decision-making in software environments > Liquid API: https://t.co/Lp1Qf15Qcc > Available on OpenRouter soon
Context
Liquid AI's d1 launch presents a model for structured decisions inside software, with a narrow overall lead that varies by task. The announcement claims it is the first model to outperform another decision model, Jev, on Hugging Face's Decision Index, but the accompanying comparison is labeled an internal reproduction of Decision Index 0.2.1.
In that company chart, d1 scores 58.9 overall against Jev 1.13's 57.9. It leads in arts, language and retrieval, while trailing in tools by 1.0 point and in knowledge by 8.0. The aggregate lead therefore supports a task-dependent comparison: the category closest to an application's decisions may matter more than the overall ranking.
- benchmark — A standard public test set for comparing AI models — the shared scoreboards behind every "model X beats model Y" claim.
Checking sign-in…
Loading comments…