Can π0.5 run on GeForce RTX 4090?

19.34 Hz inference on NVIDIA GeForce RTX 4090 (llama-cpp, fp16, 2 cam). Measured inference cost only, not task success. Sources: https://caniembody.pages.dev/compute/rtx-4090

Matrix/Compute/Consumer GPU

What can I run on GeForce RTX 4090?

24 GB VRAMAda Lovelace1321 AI TOPS450 W (total graphics power)View matrix for this target →

NVIDIA's AI TOPS headline is a low-precision, sparsity-assisted figure, not FP16/BF16 throughput. The China-market RTX 4090 D is a different, cut-down part and should not be logged against this target.

Policy by policy

Policy	Params	Inference memory (sourced, with precision)	On GeForce RTX 4090	Best measured
π0.5 VLA	—	≥8 GB	measured	19.34 Hz · llama.cpp · fp16 · 2 cam
ACT Imitation	80M	211 MB measured fp32	no data
Diffusion Policy Imitation	—	1.12 GB measured fp32	no data
EO-1 VLA	3B	≥6.5 GB	no data
Evo-1 VLA	770M	—	no data
Fast-WAM World model	6B	—	no data
GR00T N1.7 VLA	3B	≥16 GB	no data
HIL-SERL RL	—	—	no data
LingBot-VA World model	5.3B	≥18 GB	no data
MolmoAct2 VLA	—	12.1 GB measured bf16	no data
Multitask DiT Imitation	450M	—	no data
OpenVLA VLA	7.5B	15 GB measured bf16	no data
OpenVLA-OFT VLA	—	≥16 GB	no data
SmolVLA VLA	450M	1.75 GB measured fp32	no data
TD-MPC RL	—	—	no data
VLA-JEPA VLA	—	—	no data
VQ-BeT Imitation	—	—	no data
WALL-OSS VLA	—	—	no data
X-VLA VLA	900M	—	no data
π0 VLA	3.3B	13.32 GB measured fp32	no data
π0-FAST VLA	—	≥8 GB	no data

“Exceeds memory” compares a policy's upstream-stated minimum inference memory (never a measured peak, which is precision-specific) with this target's VRAM. It is arithmetic, not observation, and is labelled inferred. Quantised or split runtimes can change the answer; a measured run always wins.

All measured runs on GeForce RTX 4090

Policy	Runtime	Precision	Cams	p50 / p95 ms	Inference	Platform	Source
π0.5	llama.cpp	fp16	2 @224x224	51.7 / —	19.34 Hz	— conditions board: NVIDIA RTX 4090 · ggml_cuda_f16_f32_out: 0 (README preset for RTX 4090) Inference cost only; not task success. Full VLM + action-expert pass (ViT + LLM + 10-step action expert), LIBERO inputs (agent-view + wrist at 224x224). From the EagleVLA-Edge README RTX 4090 table (P50), PR #6 configuration. The README's naive π0.5 baseline on this GPU: 162.9 ms (not filed as a separate run: the baseline precision is not stated).	T3 · community reported
π0.5	llama.cpp	fp16	2 @224x224	58.5 / —	17.09 Hz	— conditions board: NVIDIA RTX 4090 · ggml_cuda_f16_f32_out: 0 (README preset for RTX 4090) Inference cost only; not task success. Full VLM + action-expert pass (ViT + LLM + 10-step action expert), LIBERO inputs (agent-view + wrist at 224x224). From the EagleVLA-Edge README RTX 4090 table (P50). The README's naive π0.5 baseline on this GPU: 162.9 ms (not filed as a separate run: the baseline precision is not stated).	T3 · community reported

Sources

T1 · upstream documentedvendor-docsverified 2026-10-24 (1d ago)
RTX 4090 full specs: 1321 AI TOPS (4th-gen Tensor Cores), 24 GB GDDR6X.
“Tensor Cores (AI) 4th Generation 1321 AI TOPS Boost Clock (GHz) 2.52 Base Clock (GHz) 2.23 Memory Specs: Standard Memory Config 24 GB GDDR6X”
https://www.nvidia.com/en-us/geforce/graphics-cards/40-series/rtx-4090/
T1 · upstream documentedvendor-docsverified 2026-10-24 (1d ago)
RTX 4090 full specs: total graphics power 450 W.
“Total Graphics Power (W) 450”
https://www.nvidia.com/en-us/geforce/graphics-cards/40-series/rtx-4090/
T1 · upstream documentedvendor-docsverified 2026-10-24 (1d ago)
RTX 4090 full specs: NVIDIA architecture Ada Lovelace.
“NVIDIA Architecture Ada Lovelace”
https://www.nvidia.com/en-us/geforce/graphics-cards/40-series/rtx-4090/