OpenLLaMA-3B

LLaMA family · 3.0B params

1recorded run

2.4best tok/s

1verified

Evidence: 1 sourced run · 1 independent source · first recorded Oct 4, 2026. What these labels mean.

What the OpenLLaMA-3B numbers mean

The fastest verified result on our board is 2.4 output tok/s on an GT 430 at Q4_0 with a 512-token context. That is about 416.7 ms per generated token — 500 tokens takes roughly 208.3 seconds.

How much VRAM does OpenLLaMA-3B need?

OpenLLaMA-3B has 3.0B 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)	2.7 GB	8 GB
Q8_0 (8-bit)	4.5 GB	8 GB
FP16 (16-bit)	7.7 GB	12 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)

GT 4302.4

Models with a similar footprint

Model	Params	Q4_K_M weights
LFM2.5-2.6B	3B	1.7 GB
Llama-3.2-3B-Instruct	3B	1.7 GB
Ministral-3-3B-Instruct-2512	3B	1.7 GB
VibeThinker-3B	3B	1.7 GB
gemma-3n-E4B-it-text	4B	2.2 GB
NVIDIA-Nemotron-3-Nano-4B	4B	2.2 GB

Ranked by distance from OpenLLaMA-3B's parameter count. Full benchmark board.

All benchmarks Methodology

GPU	Model	Quant	Backend	Ctx	tok/s	prompt tok/s	VRAM GB	Source
GT 430	OpenLLaMA-3B	Q4_0	llama.cpp 2e6cd4b	512	2.4 ✓	—	—	src