How long will this model take to download?
File sizes are quoted in gigabytes and connections in megabits, and the factor of eight between them is where most download estimates go wrong.
The short answer
Computed from this tool’s default settings — model size and the rest as most people start. Change them below for your own case.
A 8B model at Q4_K_M is 4.50 GB, about 8 min at 80 Mbps. Pull it once and keep it — re-downloading is usually slower than any quantisation you might have saved.
Model Download Time Calculator
File size at each quantisation, and the wait on your connection.
A 8B model at Q4_K_M is 4.50 GB, about 8 min at 80 Mbps. Pull it once and keep it — re-downloading is usually slower than any quantisation you might have saved.
| Quantisation | File size | Download time |
|---|---|---|
| Q8_0 | 8.50 GB | 14 min |
| Q6_K | 6.60 GB | 11 min |
| Q5_K_M | 5.70 GB | 10 min |
| Q4_K_M | 4.50 GB | 8 min |
| Q3_K_M | 3.50 GB | 6 min |
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 |
|---|---|---|
| Model size | 8B | 6 options, from 3B to 405B. |
| Connection speed | 100 Mbps | Megabits, as ISPs quote it. Real throughput is usually 70–85% of this. |
| Realised throughput | 80 % of line rate | Anywhere from 20 to 100 % of line rate. |
Real answers across model size
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.
| Model size | Q4_K_M download | File size | Realised speed | That is |
|---|---|---|---|---|
| 3B | 3 min | 1.69 GB | 80 Mbps | 10.0 MB/s |
| 8B | 8 min | 4.50 GB | 80 Mbps | 10.0 MB/s |
| 14B | 13 min | 7.88 GB | 80 Mbps | 10.0 MB/s |
| 32B | 30 min | 18.0 GB | 80 Mbps | 10.0 MB/s |
| 70B | 66 min | 39.4 GB | 80 Mbps | 10.0 MB/s |
A 3B model at Q4_K_M is 1.69 GB, about 3 min at 80 Mbps. Pull it once and keep it — re-downloading is usually slower than any quantisation you might have saved.
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 is parameters × bits-per-weight ÷ 8. Connection speed is quoted in megabits and file sizes are in gigabytes, so the factor of eight between them is where most estimates go wrong.
Hugging Face rarely saturates a fast line from a single connection. The realised-throughput slider is there to be honest about that rather than quoting a best case you will not see.
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.
Before pulling
Know whether this is coffee or overnight.
Slow connection
Pick a quantisation you can actually download.
Planning
Sequence a batch of downloads sensibly.
How to use this calculator
Four steps, no account, nothing leaves your browser.
Set model size
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
2 further settings: connection speed, realised throughput. 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 long will this model take to download: common questions
The questions people ask about this, answered without hedging.
How long does it take to download a model?
Divide the file size in gigabytes by your real throughput. A 4.5 GB Q4 8B model on a realistic 80 Mbps takes around eight minutes; a 40 GB 70B takes over an hour.
Why is my download slower than my connection speed?
A single connection rarely saturates a fast line, and Hugging Face throughput varies by region and time. Seventy to eighty-five percent of the line rate is a realistic expectation.
Can I resume an interrupted download?
Yes with the huggingface-cli and most clients — they resume rather than restarting. Plain browser downloads often cannot, which matters on a 40 GB file.
Should I download a smaller quant to save time?
Only if the larger one does not fit. Download time is paid once; a worse model is paid on every use.
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.