onrup

Falcon 3 · TII

Fine-tuning Falcon 3 10B

What does it take to fine-tune Falcon 3 10B?

Falcon 3 10B needs 22 GB for half-precision LoRA training and 16 GB in four-bit, and 20 GB to serve. It supports supervised fine-tuning, preference tuning (dpo), under the TII Falcon LLM licence. The cheapest qualifying class is RTX 4000 Ada at $0.09 per GPU-hour.

The top of the Falcon 3 ladder, and notable for still fitting half-precision LoRA inside a 24GB card at ten billion parameters. That makes it the largest model in the catalogue trainable on a mainstream class without quantisation.

What to know before choosing it

The interesting property is that ten billion parameters still fit half-precision LoRA inside a 24GB card, at twenty-two gigabytes. That makes this the largest model in the catalogue trainable on a mainstream class without quantisation.

For a team on a mainstream budget wanting the most capability available without moving to a 48GB class or accepting quantised training, that is a genuinely useful position, and no other model here occupies it.

Specification

Falcon 3 10B specification
Parameters10B
ArchitectureDense
Size tierMid
LoRA training memory22 GBHalf precision, frozen base
QLoRA training memory16 GBFour-bit base, higher-precision adapter
Serving memory20 GBHalf precision, before attention cache
ObjectivesSFT, DPO
LicenceTII Falcon LLM
Repositorytiiuae/Falcon3-10B-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 classVRAMTraining / hrServing / hrFits
RTX 4000 Ada20 GB$0.09$0.11QLoRA only
L424 GB$0.17$0.20LoRA and QLoRA
RTX 309024 GB$0.22$0.26LoRA and QLoRA
RTX 409024 GB$0.38$0.43LoRA and QLoRA
A4048 GB$0.42$0.48LoRA and QLoRA
RTX 6000 Ada48 GB$0.61$0.71LoRA and QLoRA
A600048 GB$0.65$0.75LoRA and QLoRA
L40S48 GB$0.78$0.90LoRA and QLoRA
A100 40 GB40 GB$1.17$1.35LoRA and QLoRA
A100 80 GB80 GB$1.68$1.94LoRA and QLoRA
H100 80 GB80 GB$2.59$2.99LoRA and QLoRA
H200141 GB$4.55$5.25LoRA and QLoRA

Half-precision LoRA needs 22 GB, so the cheapest class for it is L4 at $0.17 per hour. Below that, training has to be quantised.

Serving fits on RTX 4000 Ada at $0.11 per hour — before the attention cache, which grows with context length and concurrency.

Good starting point for

Preference tuning on this model

Preference tuning holds a frozen reference copy of the model alongside the one being trained, so budget roughly 44 GB rather than 22 GB. That is the thing that catches people out — supervised training on this model fits on a class that preference tuning will overflow.

When preference tuning beats supervised fine-tuning →

Other sizes in this family

ModelParamsQLoRAObjectives
Falcon 3 1B1B3 GBSFT, DPO
Falcon 3 3B3B6 GBSFT, DPO
Falcon 3 7B7B12 GBSFT, DPO

Comparable sizes elsewhere

Frequently asked questions

How much VRAM does it take to fine-tune Falcon 3 10B?

22 GB for half-precision LoRA and 16 GB for four-bit QLoRA. Serving needs 20 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 10B?

Four-bit QLoRA on RTX 4000 Ada at $0.09 per GPU-hour is the cheapest class that meets the 16 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 10B?

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 10B 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 10B

Upload a dataset, forecast the run, and see the cost before any compute is leased.