StarCoder2 15B Instruct on Apple M4 (32 GB)

Can the Apple M4 (32 GB) run StarCoder2 15B Instruct?

Yes, it runs well

Runs well. At Q3_K_M it needs about 10.8 GB of 21.4 GB usable, with an estimated speed of 10.3 tok/s.

Best quant: Q3_K_M
Generation speed: 10.3 tok/s
Prompt processing: 167 tok/s
First token (1K prompt): 6.1 s
Largest safe context: 16K

Best app to run it on the Apple M4 (32 GB)

On macOS

LM Studio

Desktop app · runs on llama.cpp, MLX · use the MLX build

Runs the MLX build, typically 10–20% faster than GGUF for a model with 16B active parameters on Apple silicon.
Point-and-click: search, download and chat in one window.

Open StarCoder2 15B Instruct in LM Studio

Ollama One command with Ollama’s own build of this model.
llama.cpp Usually the first to support new model architectures.

Want an agent? StarCoder2 15B Instruct supports tool use, so you can pair a llama.cpp or MLX LM server with one of these. Apps that run on this computer come first.

Coding agents

Personal assistants

How we choose: the files published for this model, your operating system and GPU, whether it fits fully in memory, and published speed tests (MLX vs llama.cpp on Apple silicon, vLLM vs llama.cpp vs Ollama on NVIDIA). Details.

Download StarCoder2 15B Instruct

Recommended for the Apple M4 (32 GB): Q3_K_M. Every other quantization has its own download link in the table below.

Quickest · Ollama

One command downloads a ready-to-run version and starts a chat.

Get Ollama at ollama.com/download (opens in a new tab) · model page (opens in a new tab)

ollama run starcoder2:15b-instruct-v0.1-q3_K_M

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

huggingface.co/lmstudio-community/starcoder2-15b-instruct-v0.1-GGUF (opens in a new tab)

ollama run hf.co/lmstudio-community/starcoder2-15b-instruct-v0.1-GGUF

Original weights · Hugging Face

Full-precision files from BigCode, for fine-tuning, vLLM, Transformers.

huggingface.co/bigcode/starcoder2-15b-instruct-v0.1 (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from BigCode. The weights themselves download from Hugging Face.

github.com/bigcode-project/starcoder2 (opens in a new tab)

git clone https://github.com/bigcode-project/starcoder2

Links checked 2 days ago against Hugging Face, Ollama and GitHub.

How it’s rated

Quality 33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking.: How capable the model itself is at Q3_K_M, from its editorial quality tier scaled by the quantization or precision. It’s the same on any hardware that runs this variant. Scale: A+ 85+ · A 70+ · B 55+ · C 40+ · D under 40.
Runs Runs well: How well it runs on the Apple M4 (32 GB): speed, capped by how well it fits in memory. Runs great: 30+ tokens/s (or near the image, video or decision speed target) · Runs well: 10+ tokens/s · Slow: 3+ tokens/s · Very slow: under 3 · Won’t run: doesn’t fit even with offload. A tight fit or RAM offload caps it at Runs well; heavy offload caps it at Slow.

Match score

To order lists, each result also gets a match score out of 100 from four parts. It favors the most capable model that still runs well.

Model qualityEditorial quality tier of the model, scaled by the quantization or precision used.	15.1 / 45
Memory fitFull marks when the model fits in the share of unified memory the GPU can use; fewer when it goes past that limit.	25.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.	7.6 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.	11.6 / 15
Match score	59.4 / 100

The match score orders results; it is not shown as a grade. Quality and Runs are the two ratings in the tables.

Every quantization on the Apple M4 (32 GB)

Quant	Memory needed	Available	Speed	Runs	Quality	Download
`Q2_K`	9.4 GB	21.4 GB	13.1 tok/s	Runs well	28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q3_K_M`	10.8 GB	21.4 GB	10.3 tok/s	Runs well	33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking.
`Q4_K_M`	12.7 GB	21.4 GB	7.9 tok/s	Slow	38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q5_K_M`	14.3 GB	21.4 GB	6.6 tok/s	Slow	38Basic quality (D) · 38 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q6_K`	16.1 GB	21.4 GB	5.5 tok/s	Slow	39Basic quality (D) · 39 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q8_0`	19.9 GB	21.4 GB	4.0 tok/s	Slowtight fit	40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`F16`	34.9 GB	21.4 GB	—	Won't run	40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 5/100 (quality, fit, speed and memory use) sets the ranking.	Original weights (opens in a new tab)

Memory figures assume 8K of context. Longer conversations or documents grow the KV cache.

Other hardware that runs it

40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 4090 Q8_0 · 39.5 tok/s · Runs great
39Basic quality (D) · 39 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. AMD Radeon RX 7900 XTX Q6_K · 44.1 tok/s · Runs great
39Basic quality (D) · 39 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 3090 Q6_K · 49.9 tok/s · Runs great
38Basic quality (D) · 38 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 4070 Ti SUPER Q5_K_M · 42.7 tok/s · Runs great
38Basic quality (D) · 38 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 4080 SUPER Q5_K_M · 46.8 tok/s · Runs great
38Basic quality (D) · 38 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 5070 Ti Q5_K_M · 56.9 tok/s · Runs great

Quick answers

How much memory does StarCoder2 15B Instruct need on the M4 32GB?

At Q3_K_M it needs about 10.8 GB. The M4 32GB has 21.4 GB usable.

How fast is StarCoder2 15B Instruct on the M4 32GB?

About 10.3 tok/s at Q3_K_M, an estimate from the hardware’s memory bandwidth and the size of the model.

Which quantization of StarCoder2 15B Instruct should I use on the M4 32GB?

Q3_K_M. Of the versions that fit, it is the one that runs best; when two run equally well, the check prefers the higher-rated one.

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.