Falcon 3 · TII
Fine-tuning Falcon 3 1B
What does it take to fine-tune Falcon 3 1B?
Falcon 3 1B needs 4 GB for half-precision LoRA training and 3 GB in four-bit, and 3 GB to serve. It supports supervised fine-tuning, preference tuning (dpo), under the TII Falcon LLM licence. The cheapest qualifying class is RTX 3080 at $0.09 per GPU-hour.
The base of the Falcon 3 ladder. Falcon 3 is unusual in how consistently the four sizes behave — a recipe tuned at 1B tends to transfer to 10B with little adjustment, which makes it a good family to prototype in when you expect to scale up later.
What to know before choosing it
The Falcon 3 ladder is unusually consistent across sizes, which makes this a good place to prototype a recipe you intend to scale. Hyperparameters tuned here transfer to the 10B with less adjustment than they would across most families.
As a base model rather than an instruction-tuned one, it has no conversational behaviour until you supply it. That is more work for a chat task and an advantage where existing instruction tuning would be something to overcome.
Specification
| Parameters | 1B |
|---|---|
| Architecture | Dense |
| Size tier | Tiny |
| LoRA training memory | 4 GBHalf precision, frozen base |
| QLoRA training memory | 3 GBFour-bit base, higher-precision adapter |
| Serving memory | 3 GBHalf precision, before attention cache |
| Objectives | SFT, DPO |
| Licence | TII Falcon LLM |
| Repository | tiiuae/Falcon3-1B-Base |
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
- Prototyping a recipe you intend to scale
- Small classification and extraction
Preference tuning on this model
Preference tuning holds a frozen reference copy of the model alongside the one being trained, so budget roughly 8 GB rather than 4 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 |
|---|---|---|---|
| Falcon 3 3B | 3B | 6 GB | SFT, DPO |
| Falcon 3 7B | 7B | 12 GB | SFT, DPO |
| Falcon 3 10B | 10B | 16 GB | SFT, DPO |
Comparable sizes elsewhere
Frequently asked questions
How much VRAM does it take to fine-tune Falcon 3 1B?
4 GB for half-precision LoRA and 3 GB for four-bit QLoRA. Serving needs 3 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 Falcon 3 1B?
Four-bit QLoRA on RTX 3080 at $0.09 per GPU-hour is the cheapest class that meets the 3 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 Falcon 3 1B?
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 Falcon 3 1B carry?
TII Falcon LLM. 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 Falcon 3 1B
Upload a dataset, forecast the run, and see the cost before any compute is leased.