Ant Group releases Ling 3.1 Flash, scoring 41 on Artificial Analysis Intelligence Index

Ant Group has released Ling 3.1 Flash, a reasoning model with 560 billion total parameters, 25 billion active parameters, and a 1 million token context window. The model is currently accessible through Novita AI at $0.30 per 1 million input tokens and $0.90 per 1 million output tokens, with cached input discounted by 80% to $0.06 per 1 million tokens. Ant Group is expected to release the model weights soon.

In benchmark testing by Artificial Analysis, Ling 3.1 Flash scored 41 on the Intelligence Index v4.3, compared to 20 for Ling 3.0 Flash. The model demonstrated gains in agentic benchmarks, recording an Elo of 1,622 on GDPval-AA v2, 1,400 on AA-Briefcase, 62% on AutomationBench-AA, and 33% on Terminal-Bench v4.0. On the AA-Omniscience test, the model scored +2, raising its accuracy rate to 29% and lowering hallucination rates to 38%.

Artificial Analysis measured Ling 3.1 Flash at a cost of $0.99 per task on the Intelligence Index. Despite its larger parameter footprint and higher unit price compared to its predecessor, it consumed 218 million output tokens to complete the index suite, running 16% fewer tokens than Ling 3.0 Flash.