How much VRAM does FLUX.2 [dev] need?
FLUX.2 [dev] needs about 32.4 GB at 8-bit, or 19.2 GB as a 4-bit GGUF, at its default output size.
- Type
- Image model
- Parameters
- 32.2B
- Text encoder
- Mistral-Small-3.1-24B-Instruct
- Released
- Nov 2025
- License
- FLUX [dev] Non-Commercial License
Memory at each precision
| Precision | Weights | Peak at 1024 × 1024 |
|---|---|---|
| 16-bit BF16 | 60.0 GB | 61.2 GB |
| 8-bit FP8 · Q8_0 | 31.1 GB | 32.4 GB |
| 6-bit GGUF Q6_K | 24.8 GB | 26.0 GB |
| 5-bit GGUF Q5_K_M | 21.4 GB | 22.6 GB |
| 4-bit GGUF Q4_K_M · NVFP4 | 18.0 GB | 19.2 GB |
The Mistral-Small-3.1-24B-Instruct text encoder (22.4 GB at FP8) runs first and then waits in system RAM.
What runs it
| Machine | Best fit | Per image |
|---|---|---|
| RTX 3060 12 GB | Too big | |
| RTX 4060 Ti 16 GB | Too big | |
| RTX 3090 | 5-bit | 4.1 min |
| RTX 4090 | 5-bit | 1.8 min |
| RTX 5090 | 6-bit | 85 s |
| RTX PRO 6000 Blackwell | 16-bit | 31 s |
| M4 Max 128 GB | 16-bit | 17 min |
| M5 Max 128 GB | 16-bit | 4.8 min |
| Ryzen AI Max+ 395 128 GB | 16-bit | 7.3 min |
| DGX Spark 128 GB | 16-bit | 4.1 min |
| M3 Ultra 256 GB | 16-bit | 9.3 min |
| M5 Ultra 512 GB | 16-bit | 2.9 min |
| H100 SXM | 16-bit | 16 s |
| DGX Station GB300 | 16-bit | 7.1 s |
1024 × 1024, 28 steps. Times are rough, plus or minus 30%.
Where to get it
Questions
How much VRAM does FLUX.2 [dev] need?
FLUX.2 [dev] needs about 32.4 GB at 8-bit, or 19.2 GB as a 4-bit GGUF, at its default output size.
Can FLUX.2 [dev] run on an RTX 4090?
Yes. At 5-bit it takes about 1.8 min per image.