RTX 4070 12 GB for local AI
Which image and video models run on the RTX 4070 12 GB, which file to download for each, and how much VRAM they need.
VRAM12 GB
MemoryGDDR6X
Bus192-bit
Bandwidth504 GB/s
ArchitectureAda Lovelace
Launched2023-04
Launch price$599
Runs well17 of 49
Specs: www.nvidia.com · launch: en.wikipedia.org
01What runs on it
| Model | Verdict | Best file | Size | Needed |
|---|---|---|---|---|
| image models | ||||
| FLUX.1 [dev] | Runs | Q5_K_S | 8.3 GB | 10.6 GB |
| FLUX.1 [schnell] | Runs | Q5_K_S | 8.3 GB | 10.6 GB |
| FLUX.1 Kontext [dev] | Runs | Q5_K_M | 8.4 GB | 11.0 GB |
| FLUX.1 Krea [dev] | Runs | Q5_K_M | 8.4 GB | 10.7 GB |
| FLUX.1 Fill [dev] | Runs | Q5_K_S | 8.3 GB | 10.9 GB |
| FLUX.2 [dev] | Offload only | Q4_K_M | 20.1 GB | 23.4 GB |
| FLUX.2 [klein] 9B | Runs well | FP8 | 9.4 GB | 11.7 GB |
| FLUX.2 [klein] 4B | Runs well | 16-bit | 7.8 GB | 9.6 GB |
| Krea 2 (Turbo) | Runs | Q5_K_M | 8.9 GB | 11.5 GB |
| Qwen-Image | Tight | Q2_K | 7.1 GB | 9.9 GB |
| Qwen-Image-Edit (2511) | Tight | Q2_K | 7.5 GB | 10.5 GB |
| Qwen-Image 2.1 | Runs well | Q8_0 | 7.6 GB | 9.9 GB |
| Z-Image Turbo | Runs well | Q8_0 | 7.2 GB | 9.2 GB |
| Z-Image (base) | Runs well | Q8_0 | 7.2 GB | 9.2 GB |
| Ideogram 4 | Offload only | Q4_1 | 6.2 GB | 14.7 GB |
| Boogu-Image (Turbo) | Runs | Q5_1 | 8.6 GB | 11.2 GB |
| ERNIE-Image (Turbo) | Runs well | Q8_0 | 8.7 GB | 11.0 GB |
| HiDream-O1-Image | Runs well | FP8 | 8.1 GB | 10.9 GB |
| Mage-Flow (Microsoft) | Runs well | 16-bit | 8.2 GB | 10.0 GB |
| Lumina Image 2.0 | Runs well | 16-bit | 5.2 GB | 7.0 GB |
| HiDream-I1 (Full) | Tight | Q3_K_M | 8.8 GB | 11.6 GB |
| HiDream-I1 (Dev) | Tight | Q3_K_M | 8.8 GB | 11.6 GB |
| Stable Diffusion 3.5 Large | Runs well | Q8_0 | 8.8 GB | 11.1 GB |
| Stable Diffusion 3.5 Medium | Runs well | 16-bit | 5.1 GB | 6.9 GB |
| Chroma1-HD | Runs well | FP8 | 9.2 GB | 11.5 GB |
| SDXL 1.0 | Runs well | 16-bit | 6.9 GB | 7.1 GB |
| Illustrious XL / Pony (SDXL anime) | Runs well | 16-bit | 6.9 GB | 7.1 GB |
| Stable Diffusion 1.5 | Runs well | 16-bit | 2.1 GB | 3.3 GB |
| HunyuanImage 2.1 | Tight | Q2_K | 7.3 GB | 10.6 GB |
| video models | ||||
| Wan 2.1 T2V 14B | Tight | Q3_K_M | 7.6 GB | 11.9 GB |
| Wan 2.1 T2V 1.3B | Runs well | 16-bit | 2.8 GB | 5.6 GB |
| Wan 2.1 I2V 14B 480P | Offload only | Q3_K_M | 8.6 GB | 12.9 GB |
| Wan 2.1 I2V 14B 720P | Offload only | Q4_K_M | 11.3 GB | 18.1 GB |
| Wan 2.1 VACE 14B | Offload only | Q3_K_S | 7.8 GB | 12.6 GB |
| Wan 2.2 T2V A14B | Tight | Q3_K_M | 7.2 GB | 11.5 GB |
| Wan 2.2 I2V A14B | Tight | Q3_K_M | 7.2 GB | 11.5 GB |
| Wan 2.2 TI2V 5B | Runs well | Q8_0 | 5.4 GB | 9.2 GB |
| Wan 2.2 Animate 14B | Tight | Q2_K | 6.5 GB | 11.8 GB |
| Wan Animate 2 (14B) | Tight | Q2_K | 6.5 GB | 11.8 GB |
| Wan 2.2 S2V 14B | Offload only | Q4_K_M | 13.9 GB | 18.7 GB |
| SCAIL-2 (character animation) | Offload only | Q2_K | 7.3 GB | 12.6 GB |
| HunyuanVideo (13B, original) | Tight | Q3_K_M | 6.2 GB | 10.5 GB |
| LTX-Video 13B (0.9.8) | Tight | Q3_K_M | 6.5 GB | 10.8 GB |
| HunyuanVideo 1.5 | Runs | Q6_K | 7.0 GB | 11.3 GB |
| LTX-2 (19B) | Offload only | Q2_K | 8.1 GB | 12.9 GB |
| LTX-2.3 (22B) | Offload only | Q2_K | 8.3 GB | 13.1 GB |
| LTX-2.5 (22B) | Offload only | Q2_K | 8.8 GB | 13.6 GB |
| MiniMax H3 (33B) | Offload only | Q4_K_M | 19.9 GB | 25.7 GB |
| MiniMax H3 Pruned | Offload only | Q4_K_M | 11.6 GB | 17.4 GB |
Calculated from real file sizes plus working memory. How the numbers work.
02Good to know
The RTX 4070 12 GB is an Ada Lovelace GPU with hardware FP8, so ComfyUI can compute Comfy-Org's FP8 files natively: small and fast. (Plain FP8 files use FP8 maths with the --fast fp8_matrix_mult option.)
03Measured and reported results
| Label | Model | Setup | Result | Peak VRAM | Date | Source |
|---|---|---|---|---|---|---|
| reported | Qwen-Image 2.1 | Qwen-Image-2.1 int8 convrot + qwen3vl_8b_int8 convrot TE + VAE bf16 · 1024x1024 · 25 steps “1024x1024・25ステップ（公式ワークフロー相当） | 14.0 秒” ComfyUI, RTX 4070 12GB with partial offload (17GB weights); same table: 12 steps 9.6 s, 2048x2048 20 steps 90.5 s; repo date not shown (Qwen-Image 2.1 era, 2026) | 14 s / image | — | — | github.com → |
| reported | Qwen-Image 2.1 | Qwen Image 2.1 INT8 ConvRot · 832x1248 · 25 steps “「全体の所要時間」で見ると、平均で32.63秒から19.83秒へ（約39.2%短縮）… GPU全体のVRAMピーク：11.19 GiB → 11.28 GiB” ComfyUI, RTX 4070 12GB, Euler/simple CFG 1; 32.63 s = standard total time (19.83 s with Spectrum speedup node); VRAM in GiB | 32.63 s / image | 11.19 GB | 2026-09-24 | note.com → |
| reported | SDXL | sd_xl_base_1.0 · 1024x1024 · 20 steps “20/20 [00:06<00:00, 3.21it/s] Prompt executed in 7.13 seconds” ComfyUI 'GPU Benchmark' thread: default workflow, SDXL 1.0 base, 1024x1024, seed 1, second run; GPU stated as 'RTX 4070 12Gb' | 7.13 s / image · 3.21 it/s | — | 2024-03-21 | github.com → |
Reported results are other people's numbers, copied as published, with a link. Settings, drivers and ComfyUI versions differ, so compare them with care. Send yours.