RTX 5070
The RTX 5070 is a cloud GPU with 12 GB of memory and 250 W TDP. It is currently available from 2 providers starting at $0.12/hr, with a market median of $0.15/hr across 2 offerings.
Hardware specifications
- Architecture
- Blackwell
- VRAM
- 12 GB
- Memory type
- GDDR7
- Memory bandwidth
- 672 GB/s
- FP32
- 31 TFLOPS
- TDP
- 250 W
Same across all providers.
Market comparison
11-day price history
Offerings
No offerings found
Try clearing the spot filter or check back later.
LLMs that fit
| Model | Params | 16-bit | 8-bit | 4-bit |
|---|---|---|---|---|
gpt-oss-120b OpenAI | 117B 5.1B active | Not released | Not released | Barely fits on 8× |
Llama 3.1 70B Meta | 70B | Needs >8 GPUs | Barely fits on 8× | Barely fits on 5× |
Qwen3.6 35B-A3B Alibaba | 35B 3B active | Needs >8 GPUs | Barely fits on 5× | Barely fits on 3× |
Qwen3 32B Alibaba | 32.8B | Barely fits on 8× | 5× · 14% spare | 3× · 6 GB free |
Qwen2.5 14B Alibaba | 14B | Barely fits on 3× | 2× · 4.7 GB free | Barely fits on 1× |
Llama 3.1 8B Meta | 8B | Barely fits on 2× | Barely fits on 1× | 1× · 5.3 GB free |
Mistral 7B Mistral AI | 7B | 2× · 5.8 GB free | 1× · 2 GB free | 1× · 6.1 GB free |
Qwen3 4B Alibaba | 4B | Barely fits on 1× | 1× · 5.6 GB free | 1× · 8.3 GB free |
Estimates: model weights plus 20% for KV cache and runtime overhead, at short context lengths. Long contexts and large batches need more. Rows marked “tight” hold the weights but not that full margin. A ×N figure is the total VRAM across N of these GPUs and makes no claim about interconnect throughput.
Similar GPUs
See more GPUs like this in
Frequently asked questions
How much does the RTX 5070 cost per hour?
The cheapest RTX 5070 offering starts at $0.12/hr, and the market median is $0.15/hr.
Which cloud providers offer the RTX 5070?
No providers currently list the RTX 5070 in our catalog.
How much VRAM does the RTX 5070 have?
The RTX 5070 has 12 GB of VRAM.
What LLMs can run on the RTX 5070?
With 12 GB of VRAM, the RTX 5070 can run 8 of our tracked models in 4-bit quantization, including gpt-oss-120b, Qwen3.6 35B-A3B, Qwen3 32B. 5 models fit in 16-bit precision.
What's the biggest LLM I can run on multiple RTX 5070s?
With 8× RTX 5070 (96 GB total), the largest tracked model that fits in 16-bit is Mistral 7B (7B params), which needs 2× GPUs.