RTX 3090 vs RTX 5070 12GB + RTX 5060 Ti 16GB x2
Best sourced output tok/s per model (single-stream generation) — engine and version, quantization, context and a source link on every row. Hover a value for the quant.
RTX 309018 sourced runs · best 709.5 tok/s
RTX 5070 12GB + RTX 5060 Ti 16GB x21 verified run · best 66.0 tok/s
Specs and evidence
| RTX 3090 | RTX 5070 12GB + RTX 5060 Ti 16GB x2 | |
|---|---|---|
| VRAM | 24 GB | 12 GB |
| Sourced runs on the board | 18 | 1 |
| Models measured | 11 | 1 |
| Best output tok/s | 709.5 tok/s | 66.0 tok/s |
| Highest measured peak VRAM | 22.0 GB | — |
| Street price (latest snapshot) | $1,720 | — |
Head-to-head
Only one model has been measured on both cards (Qwen3.8-Flash-Next). On that single data point RTX 3090 records 38.6 tok/s and RTX 5070 12GB + RTX 5060 Ti 16GB x2 records 66.0 tok/s — treat it as one data point, not a ranking.
The measured matrix
| Model | RTX 3090 | RTX 5070 12GB + RTX 5060 Ti 16GB x2 |
|---|---|---|
| LFM2.5-1.2B-Instruct | 709.5 | — |
| Qwen3.5-0.8B-MTP | 476.1 | — |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B | 353.7 | — |
| LFM2.5-2.6B | 305.6 | — |
| Qwen3.5-4B-MTP | 240.4 | — |
| Qwen3.8-27B | 164.6 | — |
| Qwen3.5-9B-MTP | 164.3 | — |
| Qwen3.8-Flash-Next | 38.6 | 66.0 |
| Ornstein3.6-27B-MTP-NSC-ACE-SABER | 32.9 | — |
| GLM-5.3-Flash | 28.1 | — |
| DeepSeek-V4.1-Flash | 26.0 | — |