30.5BTier 100Needs ~22.8 GB
Apple
Operating mode
Apple Silicon can fit a lot thanks to unified memory. This selector changes which serving posture we optimize for when surfacing the best local LLMs for this Mac.
Current mode
Balanced
Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.
Apple Silicon local AI performance. Excellent for local AI. Your Mac mini M2 24GB with 24 GB unified memory can run 77 models natively, 183 more with limits. The best match is Qwen 3.5 9B at 13 tok/s for interactive local LLM use.
77
Run great
260
Total compatible
24B
Max parameters
13
Best tok/sEST.
Comparison guide
Top models ranked for coding, chat, and writing with FAQ and buyer guidance — the comparison-intent companion to this spec sheet.
Quick picks
Top recommendations for common local AI workloads on your Mac mini M2 24GB
Mac mini M2 24GB with 24 GB unified memory. Second-generation Apple Silicon with improved GPU performance and memory bandwidth, offering a strong balance of efficiency and AI capability.
All 380 models tested
Every model ranked by how well it runs on your Mac mini M2 24GB, grouped by fit quality
These models fit comfortably and run at full speed on your Mac.
These models run but may need quantization or have reduced context windows.
These models are too large for your Mac's unified memory.
Beyond LLMs
What AI tasks this Mac can handle — from text generation to image and video creation.
| Capability | Status | Representative Model |
|---|---|---|
| LLM Chat (7B) | Runs natively | Llama 3.1 8B Q4 |
| LLM Coding (30B) | Needs offload | Qwen 3 30B Q4 |
| LLM Large (70B) | Won’t fit | Llama 3.1 70B Q4 |
| Image Gen (SDXL) | Runs natively | SDXL 1.0 FP16 |
| Image Gen (Flux) | Won't fit | Flux.1 Dev FP16 |
| Image Gen (SD 3.5) | Runs with sequential offload | SD 3.5 Large FP16 |
| Video Short (25f) | Runs natively | LTX Video 2B |
| Video Long (100f) | Won't fit | Wan Video 14B |
Same chip, more memory
Compare M2 configurations to see which models become available
16 GB unified memory
60
Run great
214
Total fit
32 GB unified memory
91
Run great
297
Total fit
Architecture
Apple M2 is the second generation of Apple Silicon, with improved GPU cores and higher memory bandwidth. The M2 Ultra scales to 192 GB unified memory via UltraFusion die-to-die interconnect.
AI Relevance
Higher memory bandwidth (~50% more than M1 in Ultra config) directly improves token generation speed for LLMs. The M2 Ultra with 192 GB unified memory can run 70B models at full Q4 quantization with good performance.
M2 brings a 10-core GPU with improved memory bandwidth. The 100 GB/s bandwidth in base models and up to 200 GB/s in Pro/Max variants provides solid decode throughput for local LLMs.
All workloads
The best local LLM for each task on your Mac mini M2 24GB
Chat
SQwen 3.5 9B matches Chat and keeps a practical fit profile. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
Coding
SQwen 3.5 9B is a specialized fit for Coding. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
Agentic Coding
SQwen 3.5 9B is a specialized fit for Agentic Coding. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
Reasoning
SQwen 3 14B matches Reasoning and keeps a practical fit profile. It is a recent-generation family, which helps on current local SOTA workloads. It should run, but memory headroom will be limited. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
RAG
AGranite 4.1 8B matches RAG and keeps a practical fit profile. It sits in the middle of the current generation mix. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama.
Just out of reach
High-quality models that need a bit more memory
Image & Video Generation
28 of 56 models can generate images or video on your Mac mini M2 24GB
| Model | Max Resolution | Gen Time | Grade |
|---|---|---|---|
| SD TurboImage | 512×512 | ~4.8s | S |
| Stable Diffusion 1.5Image | 512×768 | ~9.6s | S |
| Realistic Vision v5.1Image | 512×768 | ~9.6s | S |
| DreamShaper 8Image | 512×768 | ~9.6s | S |
| LCM DreamShaper v7Image | 512×768 | ~2.9s | S |
| PixArt-SigmaImage | 1024×1024 | ~38.4s | S |
| SDXL TurboImage | 512×512 | ~4.8s | S |
| SDXL LightningImage | 1024×1024 | ~14.4s | S |
| Stable Diffusion XL 1.0Image | 1024×1024 | ~38.4s | S |
| Playground v2.5Image | 1024×1024 | ~57.5s | S |
| RealVisXL v5.0Image | 1024×1024 | ~43.1s | S |
| DreamShaper XLImage | 1024×1024 | ~43.1s | S |
| Juggernaut XL v9Image | 1024×1024 | ~43.1s | S |
| Animagine XL 3.1Image | 1024×1024 | ~43.1s | S |
| Pony Diffusion V6 XLImage | 1024×1024 | ~43.1s | S |
| Animagine XL 4.0Image | 1024×1024 | ~43.1s | S |
| Illustrious XLImage | 1024×1024 | ~43.1s | S |
| Wan Video 2.1 1.3BVideo | 256×256 | ~28s/frame | S |
| Stable Diffusion 3.5 MediumImage | 1024×1024 | ~1m 7s | S |
| Flux.2 Klein 4BImage | 256×256 | ~25.9s | S |
| LTX Video 2BVideo | 256×256 | ~1m 40s/frame | S |
| KolorsImage | 256×256 | ~3m 24s | S |
| Stable CascadeImage | 1024×1024 | ~1m 36s | B |
| AuraFlow v0.3Image | 1024×1024 | ~2m 53s | B |
| Stable Diffusion 3.5 LargeImage | 256×256 | ~9m 30s | B |
| Stable Diffusion 3.5 Large TurboImage | 256×256 | ~1m 44s | B |
| CogVideoX 2BVideo | 256×256 | ~1m 40s/frame | D |
| Z-Image TurboImage | 256×256 | ~1m 19s | D |
| FramePack I2VVideo | 256×256 | ~1m 10s/frame | F |
| HunyuanVideoVideo | 256×256 | ~1m 10s/frame | F |
| ChromaImage | 256×256 | ~38.4s | F |
| Flux.1 DevImage | 256×256 | ~2m 53s | F |
| Flux.1 SchnellImage | 256×256 | ~33.6s | F |
| LTX Video 13BVideo | 256×256 | ~1m 10s/frame | F |
| Flux.1 Kontext DevImage | 256×256 | ~3m 12s | F |
| AnimateDiff v1.5.3Video | 512×768 | ~17.5s/frame | F |
| Cosmos Diffusion 7BVideo | 256×256 | ~1m 46s/frame | F |
| CogVideoX 5BVideo | 256×256 | ~48s/frame | F |
| Wan2.2 TI2V 5BVideo | 256×256 | ~48s/frame | F |
| Flux.2 Klein 9BImage | 256×256 | ~19.2s | F |
| Flux.1 Fill DevImage | 256×256 | ~2m 43s | F |
| Krea 2Image | 256×256 | ~52.2s | F |
| Sulphur 2Video | 256×256 | ~1m 1s/frame | F |
| Ideogram 4Image | 256×256 | ~47.2s | F |
| Mochi 1 PreviewVideo | 256×256 | ~1m 3s/frame | F |
| HunyuanVideo 1.5Video | 256×256 | ~58.8s/frame | F |
| Helios 14BVideo | 256×256 | ~1m 13s/frame | F |
| SkyReels V2 14BVideo | 256×256 | ~1m 13s/frame | F |
| Wan Video 2.1 14BVideo | 256×256 | ~1m 13s/frame | F |
| Wan Video 2.2 14BVideo | 256×256 | ~1m 13s/frame | F |
| Qwen ImageImage | 256×256 | ~1m 5s | F |
| Qwen Image EditImage | 256×256 | ~1m 5s | F |
| LTX-2 22BVideo | 256×256 | ~1m 27s/frame | F |
| Flux.2 DevImage | 256×256 | ~30m 14s | F |
| MAGI-1Video | 256×256 | ~1m 30s/frame | F |
| HunyuanImage 3.0Image | 256×256 | ~1m 54s | F |
Image models estimated at 1024×1024 (28 steps, FP16). Video models estimated at 768×512 (25 frames, 30 steps, FP16). Actual performance varies with runtime and system load.
Get started in 2 minutes
Everything you need to start running models locally with Metal acceleration and Apple Silicon unified memory
Ollama runs natively on macOS with Metal GPU acceleration. One command to install.
curl -fsSL https://ollama.com/install.sh | shQwen 3.5 9B is the best match for your Mac mini M2 24GB. Pull and run it:
ollama run qwen3.5:9bUpgrade paths
See what you unlock with more unified memory
Upgrade options
RTX 4000 Ada 20GBNext step upUnlocks 13 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 193%.
~$1,250 MSRP
Unlocks 31 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 57%.
~$1,999 MSRP
Intel Arc Pro B60 24GBBest valueUnlocks 36 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 153%.
~$599 MSRP
Unlocks 81 additional models that do not fit on the current setup.
Lifts average decode speed across fitting models by about 995%.
~$8,000 MSRP
Yes! Mac mini M2 24GB (24 GB unified memory) can run 77 models at full speed and 260 total. Top picks: Qwen 3.5 9B (score: 91/100), Qwen 3 8B (score: 89/100), Qwen 3.5 4B (score: 88/100). See the full tiered compatibility list above.
Mac mini M2 24GB has 24 GB of unified memory shared between CPU and GPU, all available for AI model inference. Unlike discrete GPUs with separate VRAM, unified memory means models can use the full 24 GB without data transfer overhead.
Not exactly. Unified memory is excellent for making larger models fit on Apple Silicon, because the CPU and GPU share one memory pool. But it is still not identical to dedicated VRAM on a high-bandwidth discrete GPU. For local AI, unified memory often wins on flexibility and capacity, while discrete GPUs can still win on raw tokens per second once a model fits comfortably.
Yes, Mac mini M2 24GB is excellent for running LLMs locally. With 24 GB unified memory and Metal acceleration, it handles 260 models with top scores above 80/100.
Because fit and speed are not the same thing. Mac mini M2 24GB can often fit larger models thanks to unified memory, but a smaller NVIDIA GPU with fast dedicated VRAM and mature CUDA kernels can still deliver higher decode throughput once the model fits. In practice, Apple Silicon is excellent for flexible local AI on one machine, while CUDA often stays ahead for the easiest setup and highest raw inference speed.
We recommend using llama.cpp on Mac mini M2 24GB. Install it with a single command, then pull your preferred model. llama.cpp supports Metal acceleration out of the box on Apple Silicon.
For coding on Mac mini M2 24GB, we recommend Qwen 3.5 9B. It achieves 9.9 tokens per second with 47K context window using 13.1 GB of unified memory. Qwen 3.5 9B is a specialized fit for Coding. It is a recent-generation family, which helps on current local SOTA workloads. It fits natively with comfortable headroom. Context coverage stays within the requested workload envelope. Known distribution channels: huggingface, ollama, lm-studio.
Mac mini M2 24GB can run Flux.1 Dev with sequential offloading or at reduced precision (FP8/NF4). The Schnell variant is faster and fits more easily in 24 GB unified memory.
Mac mini M2 24GB (24 GB unified memory) supports various AI generation tasks. For image generation, SDXL and Stable Diffusion 3.5 run well with Metal acceleration. For video, LTX Video 2.3 can generate short clips.
Mac mini M2 24GB is good for AI image generation. It handles SDXL and SD 3.5 well with Metal acceleration. Larger models like Flux may need offloading.
There are 4 upgrade path(s) from Mac mini M2 24GB: RTX 4000 Ada 20GB (20 GB), MacBook Pro M1 Pro 32GB (32 GB). Upgrading would unlock larger models like Qwen3-Coder 30B A3B Instruct and Qwen 3.5 397B A17B and faster inference.
Yes, Mac mini M2 24GB with 24 GB can run Qwen 3.5 9B at Q8 (near-lossless, ~9.6 GB) and Qwen 3.5 27B at Q4 with limited context. For best results on Mac, use MLX or Ollama.
The best local LLMs for Mac mini M2 24GB (24 GB) are: Qwen 3.5 9B (91/100, 13 tok/s), Qwen 3 8B (89/100, 14 tok/s), Qwen 3.5 4B (88/100, 29 tok/s). These models fit natively in unified memory with room for context. For coding, try the top coding pick above. For general chat, the highest-scored model gives the best Apple Silicon local AI experience.
Mac mini M2 24GB achieves 9-13 tok/s for well-fitted models with 100 GB/s memory bandwidth. Token generation speed on Apple Silicon is primarily limited by memory bandwidth and fit. Comfortable reading speed is about 6-8 tokens per second, so most natively-fitting models will feel responsive for interactive chat. MLX generally delivers 10-20% better performance than llama.cpp on newer Apple Silicon chips.
Compare with similar
Related guides