Falcon40B params2k context

Falcon 40B locally

Falcon 40B is an open-weight LLM from the Falcon family with 40B parameters. Main use: chat, RAG and general assistance. Detected minimum hardware: RTX 5090 (32 GB).

Technical facts
Parameters40B
Max context2k
Q4_K_M25.1 GB
Q5_K_M30.7 GB
Q844.7 GB
FP1689.4 GB
FamilyFalcon
Last sync2026-05-12

Available quantizations

Q4_K_M
25.1GB

Acceptable. Good compromise when VRAM is limited.

Q5_K_M
30.7GB

Good quality. Sweet spot for size and precision.

Q8
44.7GB

Near-FP16 quality. Comfortable for production.

FP16
89.4GB

Reference precision. Maximum quality, doubled VRAM.

Compatible GPUs

GPUs that can run Falcon 40B on a single card, ranked by VRAM headroom.

Recommended multi-GPU rigs

For Falcon 40B at higher quantization or with more context, a multi-GPU rig gives more headroom.

Recommended rig

4× GTX 1070

Falcon 40B with Ubuntu, vLLM, Open WebUI and the model already downloaded.

Configure

Similar models

VRAM estimates: parameters x bits/8 plus margin. Real performance varies by engine, context length and batch size.
sync: 2026-05-12