DeepSeek V4.1 Flash Launches Open-Source, Beats GPT-5.6 at One-Third Cost

DeepSeek V4.1 Flash is now open-source on Together AI. The 552B multimodal MoE reportedly outperforms GPT-5.6 Sol and Claude Opus 5 on several agentic benchmarks at about one-third the per-task cost.

2026-09-12 ~ 2026-09-14 · 4 related posts

• DeepSeek V4.1 Flash tech report: KV cache compression lets 552B beat 1.6T — 机器之心 · 2026-09-12
• DeepSeek V4.1 Flash lands on Together AI at one-third the cost per task — togethercompute · 2026-09-13
• DeepSeek V4.1 Flash: 890 bytes/token KV cache, 552B MoE beats V4 Pro on agentic tasks — togethercompute · 2026-09-13
• DeepSeek V4.1 Flash ships open weights, beats GPT-5.6 Sol and Claude Opus 5 on 4 of 5 agentic benchmarks — AryHHAry · 2026-09-14