DeepHermes-3-Mistral-24B-Preview

Mistral family · 24.0B params

1recorded run

24.5best tok/s

1verified

Evidence: 1 sourced run · 1 independent source · first recorded Jul 12, 2026. What these labels mean.

What the DeepHermes-3-Mistral-24B-Preview numbers mean

The fastest verified result on our board is 24.5 output tok/s on an RTX 3060 x2 at Q4_K_M with a 512-token context. That is about 40.7 ms per generated token — 500 tokens takes roughly 20.4 seconds. Prompt processing on the same record runs at 825.2 tok/s, so a 2,000-token prompt is read in about 2.4 seconds. Peak VRAM was measured at 19.5 GB.

How much VRAM does DeepHermes-3-Mistral-24B-Preview need?

DeepHermes-3-Mistral-24B-Preview has 24.0B 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.

Quantisation	Estimated VRAM	Smallest card that fits
Q4_K_M (4-bit)	16.3 GB	24 GB
Q8_0 (8-bit)	30.1 GB	48 GB
FP16 (16-bit)	56.0 GB	80 GB

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.

Fastest GPUs for this model (best tok/s)

RTX 3060 x224.5

Models with a similar footprint

Model	Params	Q4_K_M weights
LFM2-24B-A2B	24B	13.5 GB
mistralai_Mistral-Small-3.2-24B-Instruct-2506	24B	13.5 GB
Trinity-Mini	26B	14.6 GB
gemma-4-26B-A4B-it	27B	15.2 GB
gemma-4-26B-A4B-it-qat	27B	15.2 GB
gemma-4-26B-A4B-it-ultra-uncensored-heretic	27B	15.2 GB

Ranked by distance from DeepHermes-3-Mistral-24B-Preview's parameter count. Full benchmark board.

All benchmarks Methodology

GPU	Model	Quant	Backend	Ctx	tok/s	prompt tok/s	VRAM GB	Source
RTX 3060 x2 · 12 GB	DeepHermes-3-Mistral-24B-Preview	Q4_K_M	llama.cpp 13f2b28	512	24.5 ✓	825.2	19.5	src