Radeon AI Pro R9700 vs RTX 3080
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.
Radeon AI Pro R970014 sourced runs · best 969.4 tok/s
RTX 30809 verified runs · best 277.8 tok/s
Specs and evidence
| Radeon AI Pro R9700 | RTX 3080 | |
|---|---|---|
| VRAM | 32 GB | 10 GB |
| Sourced runs on the board | 14 | 9 |
| Models measured | 14 | 9 |
| Best output tok/s | 969.4 tok/s | 277.8 tok/s |
| Highest measured peak VRAM | 92.2 GB | 9.7 GB |
| Street price (latest snapshot) | — | — |
Head-to-head
On the 3 models measured on both cards, Radeon AI Pro R9700 leads on 2, RTX 3080 on 1. The widest gap is on Muse-Glimmer-30B: Radeon AI Pro R9700 134.6 tok/s vs RTX 3080 6.0 tok/s (Radeon AI Pro R9700 2154% faster).
The measured matrix
| Model | Radeon AI Pro R9700 | RTX 3080 |
|---|---|---|
| Qwen3.6-35B-A3B | 969.4 | — |
| Qwen3.5-0.8B | 554.5 | — |
| LFM2.5-230M | 520.0 | — |
| LFM2.5-350M | 466.0 | — |
| Qwen3.5-27B | 286.6 | — |
| LFM2.5-8B-A1B | 241.1 | 277.8 |
| gemma-4-E2B-it | 169.7 | — |
| gemma-4-E4B-it | 137.0 | — |
| Muse-Glimmer-30B | 134.6 | 6.0 |
| Ornith-1.5-9B | — | 82.8 |
| Qwen3.8-27B-Uncensored | — | 74.1 |
| Bonsai-27B | — | 72.9 |
| gemma-4-12B-it-qat | — | 71.2 |
| Ornith-1.5-35B-A3B | — | 64.4 |
| Qwen3.8-27B | 64.3 | 44.4 |
| DeepSeek-V4-Flash-0731 | 59.4 | — |
| gpt-oss-20b | — | 57.1 |
| Qwen 3.8 Flash-Next Base | 57.0 | — |
| Qwen3.8-Flash-Next | 25.6 | — |
| GLM-5.3-Flash | 17.4 | — |