Best text-to-image models to run locally on a 32GB Mac
A leaderboard of the best text-to-image models you can run locally on a Mac with 32GB of memory. Ideogram 4.0 is #1 right now, followed by Qwen-Image-2.1 and ERNIE-Image. Models are ranked on human votes and image generation tests like GenEval. Each model shows the download that fits and the command to run it.
Scores use reviewed text to image evaluations only. Compare scores within this task. Models without task measurements are excluded.
+ more
$ terminal
↗ share
detecting your OS…
Score vs memory to run
● open weights ◆ API only ─┘ best score for the memory · x: GB of RAM / VRAM for the smallest download (log)
| # | model | task score↓ | where↓ | speed↓ | sources |
|---|
| # | model | task score↓ | where↓ | speed↓ | sources |
|---|---|---|---|---|---|
| 18 |  Ideogram 4.0 · mlx-q4 | 68±2* | local |  | |
| 19 |  Qwen-Image-2.1 · int8 convrot + Qwen3-VL 8B q4_k_m (sd.cpp) | 67±2* | local |  | |
| 34 |  ERNIE-Image · gguf-ud-q4_k_m + Ministral 3B (sd.cpp, all files in memory) | 62±3 | local |   | |
| 37 | ERNIE-Image-Turbo · gguf-ud-q4_k_m + Ministral 3B (sd.cpp) | 61±3† | local | | |
| 48 |  FLUX.2 [dev] · mlx-int4 (Flux2CLI, Swift) | 59±1 | local | +1 | |
| 52 |  Krea 2 Turbo · gguf-q8_0 + Qwen3-VL 4B q4_k_m (sd.cpp) | 58±2* | local | | |
| 53 |  Z-Image (Base) · gguf-q4_k_m + Qwen3 4B q4_k_m (sd.cpp) | 58±2 | local | | |
| 62 | Z-Image-Turbo · gguf-q3_k (+ Qwen3-4B Q4_K_M text encoder) | 56±1 | local | +1 | |
| 68 |  LongCat-Image · gguf-q4_k_m + Qwen2.5-VL q8_0 (sd.cpp) | 54±4 | local | | |
| 70 | FLUX.2 [klein] 9B · gguf-q4_0 + Qwen3 8B q4_k_m (sd.cpp) | 53±2 | local | | |
| 78 | FLUX.2 [klein] 4B · bf16 + Qwen3 4B bf16 (sd.cpp) | 49±2 | local | | |
| 86 |  Ovis-Image-7B · gguf-q4_0 + Ovis 2.5 (sd.cpp) | 48±7* | local | | |
| 97 | FLUX.1 [dev] · gguf-q4_0 | 43±1 | local |  +3 | |
| 98 |  Playground v2 (1024px aesthetic) · fp16 (sd.cpp) | 43±4* | local | | |
| 101 | FLUX.1 [schnell] · mlx-4bit | 42±3* | local | | |
| 105 |  Stable Diffusion 3.5 Large · q8_0 (sd.cpp --type) | 41±1 | local | +1 | |
| 107 | FLUX.1 Kontext [dev] · gguf-q4_k_m (sd.cpp) | 40±2* | local | | |
| 113 |  SDXL-Lightning (4-step UNet) · fp16 (sd.cpp) | 40±4* | local | | |
| 114 | Stable Diffusion 3 Medium · fp16 incl. CLIPs + T5 (sd.cpp) | 39±2 | local | | |
| 122 |  SSD-1B · fp16 (sd.cpp) | 36±3* | local | | |
| 127 | Stable Diffusion XL 0.9 (base) · q8_0 (sd.cpp --type) | 35±3* | local | | |
| 128 | SDXL-Turbo · fp16 (sd.cpp) | 35±2 | local | | |
| 129 | Stable Diffusion XL 1.0 · fp16 (sd.cpp) | 35±1 | local | +5 | |
| 133 |  Dreamlike Photoreal 2.0 · q8_0 (sd.cpp --type) | 34±4* | local | | |
| 146 |  DALL·E Mega (minDALL-E) · fp32 | 31±3* | local | | |
| 149 |  Openjourney · q8_0 (sd.cpp --type) | 29±2 | local | | |
| 150 |  Stable Diffusion v1.4 · fp32 | 28±2 | local | | |
| 151 | Stable Diffusion v1.3 · fp32 | 28±4* | local | | |
| 152 | Stable Diffusion v1.1 · fp32 | 27±3 | local | | |
| 153 | Stable Diffusion v1.5 · fp32 | 27±2 | local | | |
| 154 | Stable Diffusion v1.2 · fp32 | 27±3* | local | | |
| 155 |  LCM Dreamshaper v7 · openvino-fp16 (OpenVINO GenAI) | 24±4* | local | | |
| — |  Mage-Flow · bf16 + Qwen3-VL 4B q4_k_m (sd.cpp) | not yet calibrated | local | | |
| — | Mage-Flow-Base · bf16 + Qwen3-VL 4B q4_k_m (sd.cpp) | not yet calibrated | local | | |
| — | Mage-Flow-Turbo · bf16 + Qwen3-VL 4B q4_k_m (sd.cpp) | not yet calibrated | local | |
+ 127 more don't fit your machine or budget · show them
162 of 162 catalogue models listed here · sources and licences
* new or thinly tested: ranked on too few benchmark results · † only the maker’s own numbers so far · ~RAM is estimated; full context may need more memory