Can Qianfan OCR run on the RTX PRO 6000 Blackwell?
Yes. At 8K context the best fit is F16, using about 10.1 GB of the 91.2 GB usable on the RTX PRO 6000 Blackwell. Estimated speed ceiling: ~203 tokens/sec.
+ 0.6 GB with image input. The figures on this page are for text only. Sending images also loads the vision projector (mmproj) file: F16 at 8K context then needs about 10.7 GB (still fits). Calculator with image input on.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|
| 2-bit (Q2_K) • | 2.9 GB Fits | 3.5 GB Fits | 6.9 GB Fits | 994 |
| 3-bit (Q3_K_M) • | 3.3 GB Fits | 3.9 GB Fits | 7.3 GB Fits | 797 |
| 4-bit (Q4_K_M) • | 3.8 GB Fits | 4.4 GB Fits | 7.7 GB Fits | 658 |
| 5-bit (Q5_K_M) • | 4.2 GB Fits | 4.8 GB Fits | 8.1 GB Fits | 566 |
| 6-bit (Q6_K) • | 4.6 GB Fits | 5.2 GB Fits | 8.6 GB Fits | 493 |
| 8-bit (Q8_0) • | 5.6 GB Fits | 6.2 GB Fits | 9.6 GB Fits | 381 |
| 16-bit (F16/BF16) • | 9.5 GB Fits | 10.1 GB Fits | 13.4 GB Fits | 203 |
• real GGUF file size from Reza2kn/Qianfan-OCR-GGUF. Speed is a bandwidth ceiling (1792 GB/s), not a benchmark. Methodology.
What fits instead
- Cheapest listed GPU that runs Qianfan OCR: Arc B580 (12 GB, launch MSRP $249) at F16, fully in VRAM at 8K context.
- Largest popular model that fits the RTX PRO 6000 Blackwell: gpt-oss 120B (117B, MoE) at F16, about 61.9 GB at 8K context.
"Runs" and "fits" mean the whole model in GPU memory at 8K context (4-bit or better for the model pick). Prices are US launch MSRPs, not today's street prices. "Popular" means the 60 most-downloaded models we track.
More
Qianfan OCR on every GPU and Mac · Every model that runs on the RTX PRO 6000 Blackwell · Other models that fit 64 GB VRAM · Try other settings in the calculator
Guides: Best GPU for local LLMs by budget · Can I run gpt-oss, Llama 70B or Qwen3? · Mac vs PC for local LLMs
Data: Hugging Face Hub (baidu/Qianfan-OCR), NVIDIA specs. Model data last checked 2026-11-17; refreshed hourly.