Questflow eval: MiniMax M3.1 Flash jumps from 73.8% to 94.9%, level with DeepSeek and Qwen Flash

Citing Questflow's latest eval, M3.1 Flash rises from 73.8% to 94.9%, putting it on the same ~95% line as DeepSeek and Qwen Flash models, while noting real business bottlenecks are now dirty data and flaky APIs.

澜知 AI

@LanzhiAIArchived 2 days agoOriginally posted 2026.10.25

Questflow eval: MiniMax M3.1 Flash jumps from 73.8% to 94.9%, level with DeepSeek and Qwen Flash

澜知 AI @LanzhiAIOriginally posted 2026.10.25

Open original on X

Questflow 最新评测里，MiniMax M3.1 Flash 从上一代 73.8% 飙到了 94.9%，和 DeepSeek、Qwen 的 Flash 全挤在 95% 线上。模型跑分集体刷爆，真跑业务卡死人的全成了脏数据和抽风的接口。😂 #AI #MiniMax

x.com/lanzhiai/status/2114312484459057236Archived 2026.10.26 12:06

Open original on X