Report
DeepSeek V4.1-Flash-1017 is claimed to lead open models on RSI-Inference Bench
The person who added it says it surpassed GLM-5.3 to take eighth place on the leaderboard.
TLDR
The person who added DeepSeek V4.1-Flash-1017 to RSI-Inference Bench says it ranks first among open models and eighth on the leaderboard, ahead of GLM-5.3. They list seventh place for day-0 support, with hidden tests passed rising from 35% to 62% while Pass@1 remained at 0%, and seventh for inference optimization, with a 1.23× speed-up.
Combined views
6.9K
2 Sources, first seen ago
