How much disk will my Ollama models take?
Ollama keeps every model you ever pulled until you remove it by hand, which is how a drive quietly fills with fifteen models you tried once.
The short answer
Computed from this tool’s default settings — small models (7–8b) and the rest as most people start. Change them below for your own case.
5 models at Q4_K_M come to about 40.5 GB. Ollama keeps everything you pull until removed, so audit periodically.
Ollama Storage Planner
Disk for a library of models, before you run out mid-pull.
5 models at Q4_K_M come to about 40.5 GB. Ollama keeps everything you pull until removed, so audit periodically.
| Class | Count | Each | Subtotal |
|---|---|---|---|
| Small (8B) | 3 | 4.50 GB | 13.5 GB |
| Mid (24B) | 2 | 13.5 GB | 27.0 GB |
| Large (70B) | 0 | 39.4 GB | 0.00 GB |
ollama list ollama rm <model>
What each setting changes
Every input moves the result for a reason. This is what each one does and where to find the value for your own machine.
| Setting | Default | What it changes |
|---|---|---|
| Small models (7–8B) | 3 | Your own figure, starting from 3. Change it to match what you actually run. |
| Mid models (14–32B) | 2 | Your own figure, starting from 2. Change it to match what you actually run. |
| Large models (70B+) | 0 | Your own figure, starting from 0. Change it to match what you actually run. |
| Quantisation | Q4_K_M — 4.5 bits/weight | 5 options, from Q8_0 — 8.5 bits/weight to Q3_K_M — 3.5 bits/weight. |
Real answers across quantisation
The same calculation run at a range of settings, with everything else left at its default. These are computed by the tool itself, not written by hand.
| Quantisation | Disk required | Models stored | Quantisation | Average each |
|---|---|---|---|---|
| Q8_0 — 8.5 bits/weight | 76.5 GB | 5 | Q8_0 | 15.3 GB |
| Q6_K — 6.6 bits/weight | 59.4 GB | 5 | Q6_K | 11.9 GB |
| Q5_K_M — 5.7 bits/weight | 51.3 GB | 5 | Q5_K_M | 10.3 GB |
| Q4_K_M — 4.5 bits/weight | 40.5 GB | 5 | Q4_K_M | 8.10 GB |
| Q3_K_M — 3.5 bits/weight | 31.5 GB | 5 | Q3_K_M | 6.30 GB |
5 models at Q8_0 come to about 76.5 GB. Ollama keeps everything you pull until removed, so audit periodically.
How this is calculated
No lookup tables and no invented constants. Here is the arithmetic, so you can check it against your own numbers.
A GGUF file on disk is the weights only: parameters × bits-per-weight ÷ 8. The KV cache and runtime overhead are memory costs at run time, not disk.
Ollama keeps every model you pull until you remove it explicitly, which is how a drive fills without anyone noticing. `ollama list` shows what is stored and `ollama rm` reclaims it.
These are well-founded engineering estimates, not benchmark results. Your quantisation, runtime and context length all move the real number, and usable memory is an assumption rather than a specification. See the full methodology for every assumption behind these figures.
Who this is for
The three situations that bring people to this calculation.
Running out of disk
See where the space went.
Planning a library
Budget storage before pulling a family of models.
Choosing a drive
Size an SSD for local AI work.
How to use this calculator
Four steps, no account, nothing leaves your browser.
Set small models (7–8b)
Start at the top of the panel. Every figure recalculates as you change it — there is no submit button, because watching the number move is the point.
Adjust the rest to match your setup
3 further settings: mid models (14–32b), large models (70b+), quantisation. Defaults are the common case, so change only what differs for you.
Read the headline, then the table
The large figure answers the question. The table underneath shows how the answer changes across nearby settings, which is usually where the decision actually gets made.
Check it against the method
The arithmetic is written out above. If a number looks wrong for your hardware, the assumptions are the first place to look — usable memory and quantisation are the two that vary most.
How much disk will my Ollama models take: common questions
The questions people ask about this, answered without hedging.
Where does Ollama store models?
Under ~/.ollama/models on macOS and Linux, and %USERPROFILE%\.ollama\models on Windows. The location can be moved with the OLLAMA_MODELS environment variable.
How do I free up space?
ollama list shows what is stored and ollama rm removes a model. There is no automatic cleanup, so this is a manual habit worth keeping.
Do models share layers between tags?
Yes. Ollama is content-addressed, so two tags of the same underlying weights do not store them twice — real usage is often lower than the sum of the listed sizes.
Does an SSD make inference faster?
Only at load time. Once weights are in memory, disk speed is irrelevant to generation speed — though a fast SSD makes swapping between models much less painful.
Related local AI tools
All 50 run on the same arithmetic, so answers across them agree.
More on runtime & errors
Will it fit
Context
Speed
Quantisation
Cost
Fine-tuning
Searches this page answers
Different ways of asking the same question, all resolved above.
Now find the models that fit
Sizing is only half the problem. Model Radar takes your hardware and shows which models actually run on it, ranked by what they are good at — the same arithmetic as this page, applied to every model worth running.