Can Axion 1.0 35B run on the RTX 6000 Ada?
Yes. At 8K context the best fit is Q8_0, using about 35.8 GB of the 45.6 GB usable on the RTX 6000 Ada. Estimated speed ceiling: ~232 tokens/sec.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | 128K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|---|
| 2-bit (Q2_K) | 13.2 GB Fits | 13.3 GB Fits | 13.8 GB Fits | 15.6 GB Fits | 656 |
| 3-bit (Q3_K_M) • | 16.6 GB Fits | 16.6 GB Fits | 17.1 GB Fits | 19.0 GB Fits | 515 |
| 4-bit (Q4_K_M) • | 20.7 GB Fits | 20.7 GB Fits | 21.2 GB Fits | 23.1 GB Fits | 408 |
| 5-bit (Q5_K_M) • | 24.0 GB Fits | 24.1 GB Fits | 24.5 GB Fits | 26.4 GB Fits | 349 |
| 6-bit (Q6_K) • | 27.5 GB Fits | 27.6 GB Fits | 28.1 GB Fits | 29.9 GB Fits | 303 |
| 8-bit (Q8_0) | 35.7 GB Fits | 35.8 GB Fits | 36.2 GB Fits | 38.1 GB Fits | 232 |
| 16-bit (F16/BF16) • | 65.6 GB No | 65.6 GB No | 66.1 GB No | 68.0 GB No | — |
• real GGUF file size from prithivMLmods/Axion-1.0-35B-GGUF. Speed is a bandwidth ceiling (960 GB/s), not a benchmark. Methodology.
More
Axion 1.0 35B on every GPU and Mac · Every model that runs on the RTX 6000 Ada · Other models that fit 48 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (Axion1-0/Axion-1.0-35B), NVIDIA specs. Data as of 2026-11-17.