What can a RTX 5070 Ti 16GB run?
16GB · 896 GB/s · 300W · $900
The value Blackwell: 16 GB at a hair under a terabyte-per-second keeps 32B offloads peppy.
Runs great18
Comfortable headroom — daily-driver territory.
| Model | Quant | Est. speed | Max context |
|---|---|---|---|
| deepseek moe 16b chat | FP16 | Infinity tok/s | — |
| qwen2 7b instruct | FP16 | Infinity tok/s | — |
| Qwen3-0.6B | FP16 | 447 tok/s | 33k |
| Gemma-3n-E4B | FP16 | 198 tok/s | 72k |
| SmolLM2-1.7B | FP16 | 163 tok/s | 8k |
| Llama-3.2-3B | FP16 | 86 tok/s | 91k |
| Phi-4-mini | FP16 | 72 tok/s | 73k |
| Mistral-7B-Instruct | Q8_0 | 72 tok/s | 33k |
| Gemma-3-4B | FP16 | 72 tok/s | 73k |
| Qwen3-4B | FP16 | 71 tok/s | 33k |
| DeepSeek-R1-Distill-Qwen-7B | Q8_0 | 69 tok/s | 54k |
| Qwen2.5-Coder-7B | Q8_0 | 69 tok/s | 54k |
| Llama-3.1-8B | Q8_0 | 65 tok/s | 50k |
| Qwen3-8B | Q8_0 | 65 tok/s | 33k |
| Phi-4-14B | Q4_K_M | 62 tok/s | 16k |
| DeepSeek-R1-Distill-Qwen-14B | Q4_K_M | 62 tok/s | 40k |
| Gemma-3-12B | Q5_K_M | 60 tok/s | 39k |
| Qwen3-14B | MLX 4-bit | 58 tok/s | 33k |
Tight squeeze12
It fits, but close the browser tabs and mind the context.
| Model | Quant | Est. speed | Max context |
|---|---|---|---|
| Qwen3-30B-A3B | IQ2_M | 498 tok/s | 25k |
| Qwen3-Coder-30B-A3B | IQ2_M | 498 tok/s | 25k |
| GPT-OSS-20B | Q4_K_M | 254 tok/s | 11k |
| Qwen3-32B | IQ2_M | 50 tok/s | 20k |
| DeepSeek-R1-Distill-Qwen-32B | IQ2_M | 50 tok/s | 20k |
| GLM-4-32B | IQ2_M | 50 tok/s | 20k |
| Qwen3.8 | Q3_K_M | 42 tok/s | 7k |
| Qwen3.8 27B | Q3_K_M | 42 tok/s | 7k |
| Gemma-3-27B | Q3_K_M | 42 tok/s | 7k |
| qwen35 27B | Q3_K_M | 42 tok/s | 6k |
| Codestral-22B | Q4_K_M | 41 tok/s | 5k |
| Mistral-Small-24B | MLX 4-bit | 41 tok/s | 4k |
Measured on this card0
Everything here was benchmarked on this exact card by someone who owns one — the tables above are our arithmetic; these numbers came off real machines.
Nobody has measured anything on this card yet. Paste a benchmark — it takes one keystroke and no form.