Memory at each precision
| Version | Weights | With context | Smallest Mac |
|---|---|---|---|
| 8-bit | 595 GB | 643 GB | more than any Mac |
| 6-bit | 455 GB | 491 GB | more than any Mac |
| 4-bit | 315 GB | 340 GB | 512 GB |
| 3-bit lower quality | 245 GB | 265 GB | 512 GB |
| 2-bit lower quality | 175 GB | 189 GB | 256 GB |
On every Apple chip
The smallest memory size of each chip that runs Ling 3.1 Flash, the best version that fits it, and the writing speed you can expect.
| Chip | Needs | Version | Tokens/s |
|---|---|---|---|
| M6 | no size fits | Too big | – |
| M5 Ultra | 512 GB | 4-bit Runs well | 38–68 |
| M5 Max | no size fits | Too big | – |
| M5 Pro | no size fits | Too big | – |
| M5 | no size fits | Too big | – |
| M4 Max | no size fits | Too big | – |
| M4 Pro | no size fits | Too big | – |
| M4 | no size fits | Too big | – |
| M3 Ultra | 512 GB | 4-bit Runs well | 26–46 |
| M3 Max | no size fits | Too big | – |
| M3 Pro | no size fits | Too big | – |
| M3 | no size fits | Too big | – |
| M2 Ultra | no size fits | Too big | – |
| M2 Max | no size fits | Too big | – |
| M2 Pro | no size fits | Too big | – |
| M2 | no size fits | Too big | – |
| M1 Ultra | no size fits | Too big | – |
| M1 Max | no size fits | Too big | – |
| M1 Pro | no size fits | Too big | – |
| M1 | no size fits | Too big | – |
How to run it
- In Terminal with Apple’s MLX: There is no MLX build yet, and stock mlx-lm can’t convert the official weights yet (bailing_moe_v3.py).The official weights are FP8 in 128×128 blocks; mlx-lm 0.32.2 has the architecture (bailing_moe_v3), but neither its loader nor the model's sanitize dequantizes this FP8 format, so loading rejects the weight_scale_inv tensors.
Step by step: download and run LLMs on a Mac · how to use MLX Harness · MLX vs Ollama vs llama.cpp
About Ling 3.1 Flash
| Made by | inclusionAI |
|---|---|
| Parameters | 560B total · 25B active (mixture of experts) |
| Context | 256K tokens |
| Licence | MIT |
| Released | 2026-10 |
| Input | Text |
| Intelligence | 41.1 on the Artificial Analysis Intelligence Index |
| Weights | inclusionAI/Ling-3.1-flash |
Questions
- How much memory does Ling 3.1 Flash need?
- About 315 GB of weights at 4-bit, plus room for the context: 340 GB in all. On a Mac that means 512 GB of unified memory or more, since macOS lets the GPU use about 75% of it by default.
- How fast is Ling 3.1 Flash on a Mac?
- Writing speed depends on memory bandwidth. The fastest Mac for it here is an M5 Ultra with 512 GB at about 38–68 tokens a second (4-bit). The smallest Mac that runs it, with 512 GB, manages about 38–68 on an M5 Ultra.
- Why does Ling 3.1 Flash run faster than its size suggests?
- It is a mixture-of-experts model: it holds 560B parameters in memory but reads only 25B for each token, so it writes at the speed of a much smaller model.
Similar models
For your model card or README
Shows the smallest Mac that runs Ling 3.1 Flash well and links here.
[](https://mlx.app/run/ling-3.1-flash)How it’s worked out
Memory: the model’s parameters at 8, 6, 4, 3 or 2 bits per weight plus MLX’s group scales (half a bit per weight), or its native format, with room for the context on top. By default macOS lets the GPU use about 75% of unified memory. Speed: memory bandwidth divided by the bytes read per token is the ceiling for one conversation; in llama.cpp’s public Apple Silicon benchmarks real runs reach about 45% to 80% of it, which is the range shown. These are estimates, not measurements: measure your own Mac on MLX Bench. Model figures from each official model card, chip figures from Apple’s Tech Specs. Checked 2026-10-19. Same data as mlx.app/compare.
Open data: mlx.app/data (JSON, CSV)