← Back to Model Beat
Models·Aug 15·all news from August 15, 2026

DeepSeek V4 Flash at 278 tok/s, full precision, no quantization

ModelsDeepSeek-V4-Flash

Covered by 1 source

Related stories

ModelsDeepSeek-V4-Flash-Vision-ExpAug 31 · 3 sourcesModelsShow HN: Cactus Needle 3: 8-29MB automation models can match DeepSeek V4 FlashSep 18ModelsStartLux's 27B Local Model Beats DeepSeek V4 Flash in China AI BenchmarkAug 31ModelsQwen 3.8 Flash Reduces Costs to One-Third of DeepSeek-V4-FlashAug 29