A token can't come faster than the card reads its weights, so here is the limit for every model.
1 rig on the board. All GPUs
Theoretical ceilings, not measured speeds: bandwidth divided by the weights read per token. Real runs land below them. Beside each, the best measured run on a rig with this card.
| Model | Read per token | Theoretical ceiling | Best measured | Share of its rig's ceiling |
|---|---|---|---|---|
| Qwen3.8 Flash-NextQ4_K_MReads 3.4 GB per token | 3.4 GB | 272tok/s max | Not run yet | |
| Qwen3.8 Flash-NextIQ3_SReads 2.8 GB per token | 2.8 GB | 335tok/s max | 99.0tok/s | 19% |
| Qwen3.8 Flash-NextQ8_0Reads 6.0 GB per token | 6.0 GB | 156tok/s max | Not run yet | |
| Qwen3.8 27BQ4_K_MReads 15.5 GB per token | 15.5 GB | 60.6tok/s max | Not run yet | |
| Qwen3.8 27BIQ3_SReads 11.6 GB per token | 11.6 GB | 80.6tok/s max | Not run yet | |
| Qwen3.8 27BQ8_0Reads 27.0 GB per token | 27.0 GB | 34.7tok/s max | Not run yet | |
| Llama 3.3 70BQ4_K_MReads 40.4 GB per token | 40.4 GB | 23.2tok/s max | Not run yet | |
| Llama 3.3 70BIQ3_SReads 30.4 GB per token | 30.4 GB | 30.8tok/s max | Not run yet | |
| Llama 3.3 70BQ8_0Reads 70.6 GB per token | 70.6 GB | 13.3tok/s max | Not run yet | |
| Qwen3 32BQ4_K_MReads 18.8 GB per token | 18.8 GB | 49.8tok/s max | Not run yet | |
| Qwen3 32BIQ3_SReads 14.1 GB per token | 14.1 GB | 66.4tok/s max | Not run yet | |
| Qwen3 32BQ8_0Reads 32.8 GB per token | 32.8 GB | 28.5tok/s max | Not run yet | |
| Qwen3 30B-A3BQ4_K_MReads 1.9 GB per token | 1.9 GB | 495tok/s max | Not run yet | |
| Qwen3 30B-A3BIQ3_SReads 1.4 GB per token | 1.4 GB | 660tok/s max | Not run yet | |
| Qwen3 30B-A3BQ8_0Reads 3.3 GB per token | 3.3 GB | 284tok/s max | Not run yet | |
| Gemma 3 27BQ4_K_MReads 15.7 GB per token | 15.7 GB | 59.7tok/s max | Not run yet | |
| Gemma 3 27BIQ3_SReads 11.8 GB per token | 11.8 GB | 79.4tok/s max | Not run yet | |
| Gemma 3 27BQ8_0Reads 27.4 GB per token | 27.4 GB | 34.2tok/s max | Not run yet |
Mixture-of-experts models read only their active weights, which is why they run so much faster. A rig with several cards has its own ceiling, so the share is measured against that.
| Rank | Rig | Model | Generation | Share of the ceiling | Status |
|---|---|---|---|---|---|
| 1 | CMP 170HX + RTX 3090P.I.P. Rig, @zsyonyc | Qwen3.8 Flash-NextIQ3_S on Strata | 99.0tok/s | 19% | Verified run |