DeepSeek-V4-Flash-0731

Deepseek family · 304.0B params

Evidence: 4 sourced runs · 3 awaiting review · 1 independent source · first recorded Aug 8, 2026 · last recorded Sep 4, 2026. What these labels mean.

What the DeepSeek-V4-Flash-0731 numbers mean

The fastest verified result on our board is 92.1 output tok/s on an CMP 170HX x4 at FP8 with a 487713-token context. That is about 10.9 ms per generated token — 500 tokens takes roughly 5.4 seconds. Prompt processing on the same record runs at 163277.9 tok/s, so a 2,000-token prompt is read in about 0.0 seconds. Across the 7 records we hold for this model the range is 18.0–185.6 tok/s on 6 GPUs, because quantisation, context length and the engine build all move the number.

How much VRAM does DeepSeek-V4-Flash-0731 need?

DeepSeek-V4-Flash-0731 has 304.0B parameters. These estimates use the same formula as the VRAM calculator: weights, plus about 15% for a moderate-context KV cache and runtime overhead, plus a small engine floor.

Long contexts and large batches raise the real number — read the smallest card as "possible with a reduced context". Measured throughput is on the benchmark board.

Models with a similar footprint

Ranked by distance from DeepSeek-V4-Flash-0731's parameter count. Full benchmark board.

All benchmarks Methodology