← Back to Model Beat
Models·6d ago·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

ModelsShow HN: I shrank DeepSeek V4 Flash to 57GB and it wrote a compiler on my MacAug 16ModelsDeepSeek V4 Flash Runs Test Suites 33X Cheaper Than Kimi K3Aug 11ModelsDeepSeek V4 Flash GA Costs Just $0.28 per Million TokensAug 5ModelsDeepSeek V4 Flash on a Single AMD MI300XAug 4