Best GPU for Local AI Fine-tuning

Written by Jakub Rusinowski · Last updated July 21, 2026

Ranked for fine-tuning a base model: Picking a base model to adapt on your own data with LoRA/QLoRA.

Best overall: AMD Radeon RX 7900 XTX

24 GB VRAM at 960 GB/s. It runs 87 of the models that qualify for this workload; the strongest is Qwen 3.7 35B-A3B at an estimated 164.8 tokens/sec.

The picks

	GPU	VRAM	MSRP	Models that fit	Best model it runs	Est. speed
Best overall	AMD Radeon RX 7900 XTX	24 GB	$999	87	Qwen 3.7 35B-A3B	~164.8 tok/s
Best value	AMD Radeon RX 7900 XTX	24 GB	$999	87	Qwen 3.7 35B-A3B	~164.8 tok/s
Budget pick	Intel Arc B570	10 GB	$219	60	Qwen 3 14B	~27.8 tok/s
Most memory	NVIDIA DGX Spark	128 GB	$4,699	98	Qwen 3.5 122B-A10B	~25.8 tok/s

GPU	Score	VRAM	MSRP	Models fit	Est. speed	Tok/s per watt	Cost per model
AMD Radeon RX 7900 XTX	80.6	24 GB	$999	87	~164.8	0.46	$11
NVIDIA GeForce RTX 3090	79.1	24 GB	$1,499	87	~162.2	0.46	$17
NVIDIA GeForce RTX 4090	78.4	24 GB	$1,599	87	~169.9	0.38	$18
NVIDIA GeForce RTX 5090	77.4	32 GB	$1,999	87	~233	0.41	$23
AMD Radeon RX 7900 XT	62.3	20 GB	$899	78	~28.6	0.09	$12
Intel Arc B580	51.3	12 GB	$249	61	~33	0.17	$4
AMD Radeon RX 7800 XT	50	16 GB	$499	70	~43.9	0.17	$7
AMD Radeon RX 9070	49.8	16 GB	$549	70	~44.9	0.2	$8
NVIDIA GeForce RTX 5070	49.3	12 GB	$549	61	~46.9	0.19	$9
NVIDIA GeForce RTX 5060	49.3	8 GB	$299	51	~49.5	0.34	$6
AMD Radeon RX 9070 XT	49.1	16 GB	$599	70	~49.7	0.23	$9
Intel Arc B570	47.9	10 GB	$219	60	~27.8	0.19	$4

GPUs are scored on four axes: capability (45%), speed (30%), value (20%) and efficiency (5%), each normalised across the whole ranking.
Capability means the intrinsic strength of the best model the card can hold — not how many models fit, and not how fast it streams a small one.
Speed is capped at 40 tok/s: past that, more throughput does not change how the model feels to use.
Apple Silicon is excluded from this ranking. Those entries price a whole computer and rate a chip's package power, so on price-per-capability and performance-per-watt they would beat every add-in card by construction. Apple hardware is covered on the macOS platform page instead.
Ranked as a *base model* choice, not an assistant choice: licence weight is near-maximum (0.9) because a restrictive licence blocks distributing what you train, and latency is nearly ignored (0.2) because training throughput, not decode speed, sets the cost.

The AMD Radeon RX 7900 XTX — 24 GB of VRAM runs 87 qualifying models, the strongest being Qwen 3.7.

The Intel Arc B570 at $219, which runs 60 qualifying models.

8 GB is the entry point at which a model for this workload will run at all. More memory buys a stronger model, not just a faster one.