Ant Group's Ling-3.0-flash-Fin matches MiniMax-M2.7 with half the active parameters
ArtificialAnlys · x · 2026-09-16
Ant Group's Ling team released Ling-3.0-flash-Fin, a finance-focused open-weights model built with financial institutions for research, source-checking, valuation spreadsheets, and report writing.
- Specs: 124B total / 5.1B active parameters (MoE), 256K context, text-only, MIT license, free rate-limited endpoint on OpenRouter.
- Benchmarks: Scores 23 on the AA Intelligence Index, matching MiniMax-M2.7 with roughly half the active parameters; 24 on the Finance & Accounting Index, with higher business knowledge accuracy than Ling-3.0-flash-VL (17% vs 11%) but worse hallucination (33% vs 19%).
- Weak spots: Agentic tasks lag — 7% on AutomationBench-AA (vs 16% for VL) and 0% on Terminal-Bench v4.0; it's also verbose, averaging 67k output tokens per task, 34% more than VL and 3.2x MiniMax-M2.7.
More from Models
- Users report ChatGPT sessions getting muddled, answering questions from other chats — koltregaskes · 2026-09-16
- OpenAI reportedly prepping Codex Replay to run and compare historical task threads in parallel — testingcatalog · 2026-09-16
- Mystery stealth model Union Alpha hits OpenRouter: free, 256K context, agentic focus — gaganghotra_ · 2026-09-16
- Developer builds demo hours after getting Jev access, drawing researcher banter — suchenzang · 2026-09-16
- TabPFN-3.5 tops Kaggle's Otto competition out of the box, but experts call the benchmark flawed — RichmanRonald · 2026-09-16
- Bindu Reddy teases Opus 5.2 in testing, Grok 4.8 weeks away, OpenAI's Astra+ in testing — bindureddy · 2026-09-16