Best AI Models for iMac 24" M3 (24 GB)
24 GB unified − 3.5 GB OS overhead = 20.5 GB available for AI models
How does this device compare with your machine?
37 of 293 featured models run great on iMac 24" M3 (24 GB) · best pick: Qwen3.8 27B at ~4.1 tok/s
largest that fits: 36.0B (Qwen3.6 35B A3B, IQ4_XS) · 70B needs CPU offload or more memory
What LLMs Can iMac 24" M3 (24 GB) Run?
1068 models · 229 run great · 224 run well
Score: model quality, 100 = average model. Fit: how well it runs on this hardware, S best to F won't fit. Ordered by quality, recency and Fit. How ranking works
Showing compatibility for iMac 24" M3 (24 GB)
| Model | Quant | VRAM | Speed | Context | Score | Fit |
|---|---|---|---|---|---|---|
| Q4_K_M | 10.7 GB | 12.8 t/s | 1073K | B52, Decent | ||
Qwen2.5 Coder 14B InstructChatCodeFunctions | Q4_K_M | 12.4 GB | 7.5 t/s | 33K | B45, Decent | |
Llamatron 8B V1Chat | Q4_K_M | 7.3 GB | 13.3 t/s | 131K | B53, Decent | |
| BF16 | 1.2 GB | 68.9 t/s | 131K | S86, Runs great | ||
| BF16 | 2.5 GB | 55.1 t/s | — | A82, Runs well | ||
Mamba 130M HFChat | Q4_K_M | 0.9 GB | 796.4 t/s | — | S100, Runs great | |
XiaoHong V1Chat | Q4_K_M | 7.6 GB | 13.0 t/s | 41K | B52, Decent | |
MobileLLM R1.5 950MChatReasoning | BF16 | 3.5 GB | 34.3 t/s | — | A73, Runs well | |
InternVL2 2BVision | BF16 | 6.6 GB | 14.8 t/s | — | B55, Decent | |
| Q4_K_M | 2.2 GB | 101.0 t/s | — | S92, Runs great | ||
Qehwa Pashto LlmChat | Q4_K_M | 5.8 GB | 14.4 t/s | 131K | B54, Decent | |
Schematron 3BChat | Q4_K_M | 4.1 GB | 29.1 t/s | 131K | A70, Runs well | |
| Q4_K_M | 3.6 GB | 41.4 t/s | — | A76, Runs well | ||
| Q4_K_M | 3.4 GB | 45.1 t/s | 33K | A77, Runs well | ||
| Q4_K_M | 11.8 GB | 7.5 t/s | 41K | B45, Decent | ||
Medgemma 1.5 4B ITVision | Q4_K_M | 5 GB | 25.2 t/s | — | A66, Runs well |
iMac 24" M3 (24 GB) has 20.5 GB of memory for models. In the ranking below, Qwen3.8 27B leads at Q4_K_M: it needs 17.9 GB (87% of that memory) and generates about 4.1 tokens per second. 37 of the 293 featured models earn a Fit S here, meaning they load with room for context and run fast. Decode speed is set mostly by memory bandwidth (102 GB/s), so smaller and mixture-of-experts models are noticeably quicker than dense models of the same quality. 70B-class models need about 46 GB at Q4_K_M, more than the 20.5 GB available, so they require offloading layers to system RAM or a larger-memory setup.
Runs Well
- GPT OSS 20B (20.9B at Q4_K_M, ~24.4 tok/s)
- Ling 3.0 Tiny (7.9B at Q4_K_M, ~56.1 tok/s)
- MiniCPM5 2B (2.5B at Q4_K_M, ~37.7 tok/s)
- Qwen3.5 2B (2.3B at Q4_K_M, ~40.5 tok/s)
- LFM2.5 8B A1B (8.5B at Q4_K_M, ~50 tok/s)
Challenging
- Qwen3.8 27B (27.8B at Q4_K_M, ~4.1 tok/s) fits but uses 87% of memory, so keep context short
- Gemma 4 26B A4B IT (25.8B at Q4_K_M, ~22.6 tok/s) fits but uses 80% of memory, so keep context short
- Qwen3.6 35B A3B (36.0B at IQ4_XS, ~22.6 tok/s) fits but uses 97% of memory, so keep context short
- 70B-class models need ~46 GB at Q4_K_M, so they need CPU offload or more memory
- 120B-class models need ~77 GB at Q4_K_M, so they need CPU offload or more memory
iMac 24" M3 (24 GB) Specifications
- Brand
- Apple
- Chip
- M3
- Type
- Desktop
- Unified Memory
- 24 GB
- Memory Bandwidth
- 102.4 GB/s
- GPU Cores
- 10
- CPU Cores
- 8
- Neural Engine
- 18.0 TOPS
- Release Date
- 2023-11-07
Get Started
Prompt Processing
Estimated for Qwen3.8 27B, the compute-bound phase that reads your prompt before the first reply token appears.
Short chat
20.3 s
512 tok prompt
Long chat
163 s
4,096 tok prompt
Document / codebase
1300 s
32,768 tok prompt
Prefill is compute-bound and a different number from the decode tok/s shown elsewhere on this page — how prompt processing works →
Performance figures are estimates calibrated as of 2026-09-21 — see calibration basis →
What to do next with iMac 24" M3 (24 GB)
Which model to run first, a faster option with the same memory, and what more memory unlocks. Computed from the rankings and specs above.
Best coding model here
Qwen3.8 27B
27.8B at Q4_K_M
- Coding llmrun Score 112.1
- about 4.1 tokens per second
- 17.9 GB needed at Q4_K_M
Highest coding llmrun Score among official models that fit iMac 24" M3 (24 GB).
Same 24 GB, faster
MacBook Pro 14-inch (M5)
Apple
- 8 GB more memory
- 50% higher memory bandwidth
Higher memory bandwidth means faster token generation.
Bigger models: step up to 32 GB
Mac Mini M4 (32 GB)
Apple
- 8 GB more memory
- 17% higher memory bandwidth
Unlocks 8 more official models, incl. Mixtral 8x7B Instruct v0.1 (46.7B at IQ4_XS).
Frequently Asked Questions
Can iMac 24" M3 (24 GB) run Qwen3.8 27B?
Yes. The iMac 24" M3 (24 GB) with 24 GB unified memory can run Qwen3.8 27B (Q4_K_M, 17.9 GB, 87% of memory). It is also the table's best pick, at ~4.1 tok/s. 37 of the 293 featured models get a Fit of S (runs great) here. The largest featured model that fits at Q4 or better is Qwen3.6 35B A3B (36.0B at IQ4_XS). 70B-class models need around 46 GB at Q4_K_M, meaning they need CPU offloading or extra memory beyond what's available in 20.5 GB. Check the compatibility table above for the full list with Fit, VRAM usage and estimated speed. Models are checked against the 20.5 GB available to AI (that is, the 24 GB of RAM minus roughly 3.5 GB for macOS), while Apple Silicon's unified memory allows the GPU to draw from that same pool without copying data.
How much memory is available for AI on iMac 24" M3 (24 GB)?
The iMac 24" M3 (24 GB) has 24 GB unified memory. After macOS reserves ~3.5 GB for the operating system, approximately 20.5 GB is available for AI models. Model rankings here use that 20.5 GB limit instead of the total 24 GB. Unlike discrete GPUs where VRAM is separate from system RAM, a unified-memory device shares one memory pool between the CPU and GPU — this means no data copying overhead, but you share memory with macOS and open apps.
Is iMac 24" M3 (24 GB) good for AI?
With 24 GB unified memory and 102.4 GB/s bandwidth, the iMac 24" M3 (24 GB) is solid for running local AI models. Models that run well include GPT OSS 20B (20.9B at Q4_K_M, ~24.4 tok/s); Ling 3.0 Tiny (7.9B at Q4_K_M, ~56.1 tok/s); MiniCPM5 2B (2.5B at Q4_K_M, ~37.7 tok/s). Mid-size models run at good quality, and Metal acceleration keeps interactive chat smooth.
What's the best model for iMac 24" M3 (24 GB)?
For the iMac 24" M3 (24 GB), the best pick right now is Qwen3.8 27B at ~4.1 tok/s. Also running well: GPT OSS 20B (20.9B at Q4_K_M, ~24.4 tok/s); Ling 3.0 Tiny (7.9B at Q4_K_M, ~56.1 tok/s); MiniCPM5 2B (2.5B at Q4_K_M, ~37.7 tok/s). The best choice depends on your use case: coding assistants benefit from code-tuned models, general chat from instruction-tuned ones. The list is ranked by quality, recency and Fit.
How fast is iMac 24" M3 (24 GB) for AI inference?
With 102.4 GB/s memory bandwidth, the iMac 24" M3 (24 GB) achieves approximately 16 tok/s on a 7B model at Q4_K_M — that's functional for interactive use. A 14B model runs at ~8 tok/s. In the ranking, the top pick Qwen3.8 27B runs at ~4.1 tok/s. Apple Silicon achieves high efficiency (~70%) thanks to unified memory — there's no PCIe bottleneck between CPU and GPU.
Tokens per second (estimate): Our tok/s is the decode speed for one user, estimated from memory bandwidth divided by the model size, times an efficiency factor for the platform. Real results typically land within ±20%. It does not cover prompt processing, which is measured separately as time to first token. Read more = (102.4 GB/s ÷ model GB) × efficiency
Apple Silicon achieves ~70% bandwidth efficiency thanks to Unified memory: Apple Silicon Macs pool their memory between CPU and GPU. We subtract the 3.5 GB reserved for macOS from the total RAM to find what's available for a model; thus, a 32 GB Mac counts as having 28.5 GB usable. Other unified-memory devices expose a lower fixed limit instead, which we apply. Read more and Metal acceleration.
Estimated speed on iMac 24" M3 (24 GB)
Real-world results typically within ±20%.
Can I run AI offline on iMac 24" M3 (24 GB)?
Yes — once you download a model, it runs entirely on the iMac 24" M3 (24 GB) without internet. Applications like Ollama and LM Studio make it straightforward to download, manage, and run models locally. All your conversations stay private on your device with zero data sent to external servers. This is one of the key advantages of local AI: complete privacy, no API costs, and no rate limits.