The efficiency-first member of the V4 line: MoE, ~284B total and only ~13B active per token, context up to 1M. Tuned for fast, cheap inference — everyday chat and light tasks.
The efficiency-first member of the V4 line: MoE, ~284B total and only ~13B active per token, context up to 1M. Tuned for fast, cheap inference — everyday chat and light tasks.
The efficiency-first member of the V4 line: MoE, ~284B total and only ~13B active per token, context up to 1M. Tuned for fast, cheap inference — everyday chat and light tasks.
DeepSeek V4.1 Flash 1017, the 2026-10-17 release tuned for DeepSeek Harness. Uses the DeepSeek V4 Flash tariff; this model consumes 2x DeepSeek balance compared with other DeepSeek models.