Head to head · October 2026
RTX 3090 vs Arc Pro B60: which is better for local AI?
24GB against 24GB, 936 against 456 GB/s. What each runs, how fast, and which one to buy.
The verdict
Both have 24GB of memory, so they run the same models. The RTX 3090 generates text about 160% faster thanks to 936 GB/s of memory bandwidth versus 456 GB/s. The Arc Pro B60 costs about $400 less ($650 vs $1,050 used).
- Get the RTX 3090 if speed matters most or you want the widest software support (CUDA).
- Get the Arc Pro B60 if you want the most for your money, you care about power draw, heat, and noise, or you want a new card with a warranty.
We may earn a commission when you buy through links on this page. It never changes what we recommend. How we make money
Performance
Model by model
Estimated single-user generation speed for popular models, at the best quantization each card can hold with 8K context.
| Model | RTX 3090 | Arc Pro B60 |
|---|---|---|
| Llama 3.1 8B Instruct | ~73 tok/sQ8_0 | ~28 tok/sQ8_0 |
| Qwen3 14B | ~41 tok/sQ8_0 | ~15 tok/sQ8_0 |
| gpt-oss-20b | ~120 tok/sMXFP4 | ~48 tok/sMXFP4 |
| Qwen3 30B-A3B (2507) | ~135 tok/sQ4_K_M | ~54 tok/sQ4_K_M |
| Gemma 3 27B | ~39 tok/sQ4_K_M | ~15 tok/sQ4_K_M |
| Qwen3 32B | ~41 tok/sQ3_K_M | ~15 tok/sQ3_K_M |
| Llama 3.3 70B Instruct | ~1.8 tok/sQ4_K_M Partial offload | ~1.7 tok/sQ4_K_M Partial offload |
| gpt-oss-120b | ~21 tok/sMXFP4 Partial offload | ~18 tok/sMXFP4 Partial offload |
Estimates from our bandwidth model; see methodology. Measured results, where we have them, are on each GPU's page.
Specifications
Spec by spec
| Spec | RTX 3090 | Arc Pro B60 |
|---|---|---|
| Memory | 24GB GDDR6X | 24GB GDDR6 |
| Bandwidth | 936 GB/s● | 456 GB/s |
| Power | 350W | 200W● |
| Price | ~$1,050 used | ~$650● |
| Price per GB | $44 | $27● |
| Architecture | Ampere | Xe2 (Battlemage) |
| Released | 2020 | 2025 |
| Software | CUDA, Vulkan | SYCL, Vulkan |
Our take
Bottom line on each
The default value pick for 24GB of CUDA memory and the backbone of most multi-GPU builds.
24GB for about $650 new, among the lowest prices per GB of any current card, with less polished software than CUDA.
Frequently asked questions
Is the RTX 3090 or Arc Pro B60 better for running LLMs?
Both have 24GB of memory, so they run the same models. The RTX 3090 generates text about 160% faster thanks to 936 GB/s of memory bandwidth versus 456 GB/s. The Arc Pro B60 costs about $400 less ($650 vs $1,050 used).
How much faster is the RTX 3090 than the Arc Pro B60 for AI?
About 160% faster at generating text in our estimates, because token generation is limited by memory bandwidth (936 GB/s vs 456 GB/s). Prompt processing depends more on compute and can differ by more.
Keep comparing