Ant Group's LingDT-Decision-1.0 Cuts Token Use 97%, Tops Decision-Model Benchmarks
智东西 · wechat · 2026-10-22
Ant Group's fintech arm released LingDT-Decision-1.0 (4B/27B), a decision model for customer service, risk control, and finance, now free to try on its DTMaaS platform. Decision models skip long-form generation, directly returning options, scores, or judgments for intent recognition, ticket routing, and tool selection.
Unlike the more general-purpose Jev, LingDT targets complex enterprise decisions, debuting native multi-select (handling refund + subscription + invoice requests at once) and sequential DAG decision-making based on task dependencies. Benchmarks: vs DeepSeek-V4-flash, 97% lower token use and 10x faster; vs Qwen-3.7-flash, 90% fewer tokens in customer service with risk accuracy up from 76% to 89%.
Across 23 open benchmarks, the 27B version ranked first overall, beating Microsoft-Decision-1 and Jev 1.13.0.
More from Models
- ChatGPT web can actually listen to audio, but the same model can't via API or Codex — Possible_Statement84 · 2026-10-23
- MiniMax M3.1 Flash free for 3 days on 0G, with 1M context and 512K output — MiniMax_AI · 2026-10-23
- TLA+ expert corrects Grok: TLC only exhausts finite state spaces, not all behaviors — tianyin_xu · 2026-10-23
- OpenAI blocks manual reset with 13% usage left: 'no reset needed right now' — AccomplishedSugar490 · 2026-10-23
- Meta's 816B Muse Spark Open scores 39, losing to 180B Qwen3.8-Flash-Next at 40 — BTA_Labs · 2026-10-23
- Gemini refuses to generate image in Docs, blames document size limit — viksit · 2026-10-23