Phi · Microsoft
Fine-tuning Phi-4 Mini 3.8B
What does it take to fine-tune Phi-4 Mini 3.8B?
Phi-4 Mini 3.8B needs 12 GB for half-precision LoRA training and 8 GB in four-bit, and 9 GB to serve. It supports supervised fine-tuning, preference tuning (dpo), under the MIT licence. The cheapest qualifying class is RTX 3080 at $0.09 per GPU-hour.
MIT-licensed, which is as unencumbered as it gets — no acceptable-use policy, no revenue cap, no attribution clause. The Phi training approach favours curated data over volume, and the result punches above 3.8B on reasoning and code in particular.
What to know before choosing it
MIT is as unencumbered as model licensing gets: no acceptable-use policy, no revenue cap, no attribution clause, nothing that follows the derivative. For products that ship models onward, that removes a whole category of problem.
The Phi training approach favours curated data over raw volume, and the visible effect is that code and mathematical reasoning punch well above 3.8B while general world knowledge does not. Choose it for structured reasoning; choose something else if the task needs breadth of fact.
Specification
| Parameters | 3.8B |
|---|---|
| Architecture | Dense |
| Size tier | Small |
| LoRA training memory | 12 GBHalf precision, frozen base |
| QLoRA training memory | 8 GBFour-bit base, higher-precision adapter |
| Serving memory | 9 GBHalf precision, before attention cache |
| Objectives | SFT, DPO |
| Licence | MIT |
| Repository | microsoft/Phi-4-mini-instruct |
What it costs to train
Every class with enough memory for four-bit training, cheapest first. Total cost is the rate multiplied by wall time, so the cheapest rate is not always the cheapest run — a faster class that finishes sooner frequently wins.
| GPU class | VRAM | Training / hr | Serving / hr | Fits |
|---|---|---|---|---|
| RTX 3080 | 12 GB | $0.09 | $0.11 | LoRA and QLoRA |
| RTX 4000 Ada | 20 GB | $0.09 | $0.11 | LoRA and QLoRA |
| L4 | 24 GB | $0.17 | $0.20 | LoRA and QLoRA |
| RTX 3090 | 24 GB | $0.22 | $0.26 | LoRA and QLoRA |
| RTX 4090 | 24 GB | $0.38 | $0.43 | LoRA and QLoRA |
| A40 | 48 GB | $0.42 | $0.48 | LoRA and QLoRA |
| RTX 6000 Ada | 48 GB | $0.61 | $0.71 | LoRA and QLoRA |
| A6000 | 48 GB | $0.65 | $0.75 | LoRA and QLoRA |
| L40S | 48 GB | $0.78 | $0.90 | LoRA and QLoRA |
| A100 40 GB | 40 GB | $1.17 | $1.35 | LoRA and QLoRA |
| A100 80 GB | 80 GB | $1.68 | $1.94 | LoRA and QLoRA |
| H100 80 GB | 80 GB | $2.59 | $2.99 | LoRA and QLoRA |
| H200 | 141 GB | $4.55 | $5.25 | LoRA and QLoRA |
Serving fits on RTX 3080 at $0.11 per hour — before the attention cache, which grows with context length and concurrency.
Good starting point for
- Code and maths tasks at small scale
- Products where any licence restriction is a blocker
- Reasoning on a tight budget
Preference tuning on this model
Preference tuning holds a frozen reference copy of the model alongside the one being trained, so budget roughly 24 GB rather than 12 GB. That is the thing that catches people out — supervised training on this model fits on a class that preference tuning will overflow.
Other sizes in this family
| Model | Params | QLoRA | Objectives |
|---|---|---|---|
| Phi-4 14B | 14B | 20 GB | SFT, DPO |
Comparable sizes elsewhere
Frequently asked questions
How much VRAM does it take to fine-tune Phi-4 Mini 3.8B?
12 GB for half-precision LoRA and 8 GB for four-bit QLoRA. Serving needs 9 GB. Preference tuning roughly doubles the training figure, because a frozen reference model is held alongside the one being trained.
What is the cheapest way to fine-tune Phi-4 Mini 3.8B?
Four-bit QLoRA on RTX 3080 at $0.09 per GPU-hour is the cheapest class that meets the 8 GB threshold. Note that quantised training is slower per step, so a faster class sometimes costs less over the whole run.
Can I download the weights after fine-tuning Phi-4 Mini 3.8B?
Yes. Every finished run exposes its trained weights for download, and publishing to a model hub is a single call with a generated model card recording the base model and version the adapter applies to.
What licence does Phi-4 Mini 3.8B carry?
MIT. The licence follows the fine-tune — a derivative inherits the base model’s terms, and those terms pass to anyone you give the model to.
Last verified 6 August 2026. Memory thresholds are the platform's own admission limits.
Fine-tune Phi-4 Mini 3.8B
Upload a dataset, forecast the run, and see the cost before any compute is leased.