Written by Jakub Rusinowski · Last updated July 21, 2026
Ranked for mathematics and quantitative work: Symbolic and numerical problem solving, proofs, and step-by-step quantitative reasoning.
Best overall: Intel Arc B580
12 GB VRAM at 456 GB/s. It runs 50 of the models that qualify for this workload; the strongest is Qwen 3 14B at an estimated 33 tokens/sec.
| GPU | VRAM | MSRP | Models that fit | Best model it runs | Est. speed | |
|---|---|---|---|---|---|---|
| Best overall | Intel Arc B580 | 12 GB | $249 | 50 | Qwen 3 14B | ~33 tok/s |
| Best value | Intel Arc B570 | 10 GB | $219 | 49 | Qwen 3 14B | ~27.8 tok/s |
| Budget pick | Intel Arc B570 | 10 GB | $219 | 49 | Qwen 3 14B | ~27.8 tok/s |
| Most memory | NVIDIA DGX Spark | 128 GB | $4,699 | 87 | Qwen 3 32B | ~9.6 tok/s |
| GPU | Score | VRAM | MSRP | Models fit | Est. speed | Tok/s per watt | Cost per model |
|---|---|---|---|---|---|---|---|
| Intel Arc B580 | 69.9 | 12 GB | $249 | 50 | ~33 | 0.17 | $5 |
| NVIDIA RTX 6000 Ada Generation | 69.7 | 48 GB | $6,799 | 79 | ~32 | 0.11 | $86 |
| Intel Arc B570 | 68.3 | 10 GB | $219 | 49 | ~27.8 | 0.19 | $4 |
| NVIDIA L40S | 66.7 | 48 GB | $7,499 | 79 | ~29 | 0.08 | $95 |
| AMD Radeon RX 7800 XT | 66.3 | 16 GB | $499 | 57 | ~43.9 | 0.17 | $9 |
| AMD Radeon RX 9070 | 66 | 16 GB | $549 | 57 | ~44.9 | 0.2 | $10 |
| NVIDIA GeForce RTX 5090 | 65.7 | 32 GB | $1,999 | 74 | ~57.4 | 0.1 | $27 |
| NVIDIA GeForce RTX 5070 | 65.6 | 12 GB | $549 | 50 | ~46.9 | 0.19 | $11 |
| AMD Radeon RX 9070 XT | 65.3 | 16 GB | $599 | 57 | ~49.7 | 0.23 | $11 |
| NVIDIA GeForce RTX 3080 (10GB) | 63.2 | 10 GB | $699 | 49 | ~52.4 | 0.16 | $14 |
| NVIDIA GeForce RTX 5070 Ti | 62.9 | 16 GB | $749 | 57 | ~60.5 | 0.2 | $13 |
| NVIDIA GeForce RTX 4070 Ti Super | 62.7 | 16 GB | $799 | 57 | ~46.9 | 0.16 | $14 |
The Intel Arc B580 — 12 GB of VRAM runs 50 qualifying models, the strongest being Qwen 3.
The Intel Arc B570 at $219, which runs 49 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.