NVIDIA GeForce RTX 4070
12 GB · 504.2 GB/s · 200 W
In short
The NVIDIA GeForce RTX 4070 has 12 GB of GDDR6X memory with 504.2 GB/s of bandwidth, which sets how fast it generates text. At 8k context it runs 34 of the 125 open models we track fully in memory; the most capable is Gemma 4 12B at Q6_K, at about 31 tok/s (estimated). We have no current price for the NVIDIA GeForce RTX 4070 in the US.
Updated . Speeds are estimates; how they are worked out.
Prices
Every retailer we track in the US, cheapest first. Buy links may earn us a commission.
No retailer in this market lists it right now.
What it runs
Every model in the catalogue, with the best quant for this hardware. Estimates, not measurements.
| Model | Quant | Fits | Memory | Max context | Decode @8k | Prefill | Good for |
|---|---|---|---|---|---|---|---|
| Gemma 4 12B | Q6_K | fits | 10.3 GB | 59k100k with q8 cache | 31tok/sest. | 522 tok/s | chat, coding, long documents |
| Qwen 3.5 9B Alibaba Qwen | Q8_0 | fits | 9.9 GB | 47k79k with q8 cache | 32tok/sest. | 661 tok/s | chat, coding |
| Qwen3 14B Alibaba Qwen | Q4_K_M | fits | 10.4 GB | 14k26k with q8 cache | 30tok/sest. | 399 tok/s | chat |
| Ministral 3 14B Mistral AI | Q5_K_M | fits | 11.0 GB | 10k19k with q8 cache | 29tok/sest. | 429 tok/s | chat |
| DeepSeek R1 0528 Qwen3 8B DeepSeek | Q8_0 | fits | 10.0 GB | 17k32k with q8 cache | 32tok/sest. | 698 tok/s | chat |
| Qwen3 8B Alibaba Qwen | Q8_0 | fits | 10.0 GB | 17k32k with q8 cache | 32tok/sest. | 698 tok/s | chat |
| Phi-4 14B Microsoft | Q4_K_M | fits | 10.6 GB | 12k16k with q8 cache | 30tok/sest. | 401 tok/s | chat |
| Qwen 3.5 4B Alibaba Qwen | Q8_0 | fits | 5.2 GB | 174k256k with q8 cache | 65tok/sest. | 1331 tok/s | chat, coding, long documents |
| Ministral 3 8B Mistral AI | Q8_0 | fits | 10.2 GB | 16k29k with q8 cache | 31tok/sest. | 652 tok/s | chat |
| Granite 4.2 8B IBM | Q8_0 | fits | 10.7 GB | 12k22k with q8 cache | 29tok/sest. | 647 tok/s | chat |
| Gemma 4 E4B | Q8_0 | fits | 8.6 GB | 91k128k with q8 cache | 37tok/sest. | 801 tok/s | chat, coding, long documents |
| LFM2.5 8B-A1B Liquid AI | Q8_0 | fits | 9.5 GB | 43k75k with q8 cache | 128tok/sest. | 4675 tok/s | chat, coding |
| Llama 3.1 8B Meta | Q6_K | fits | 7.9 GB | 34k63k with q8 cache | 41tok/sest. | 720 tok/s | chat, coding |
| Qwen2.5 0.5B Alibaba Qwen | Q8_0 | fits | 1.3 GB | 32k | 495tok/sest. | 9785 tok/s | chat, coding, long documents |
| Qwen3.5 0.8B Alibaba Qwen | Q8_0 | fits | 1.6 GB | 256k | 336tok/sest. | 6732 tok/s | chat, coding, agents, long documents |
| Qwen3.5 2B Alibaba Qwen | Q5_K_M | fits | 2.2 GB | 256k | 201tok/sest. | 2762 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Qwen 1.5B DeepSeek | Q8_0 | fits | 2.7 GB | 128k | 147tok/sest. | 3080 tok/s | chat, coding, long documents |
| MiniCPM5 2B openbmb | Q8_0 | fits | 3.6 GB | 128k | 103tok/sest. | 2036 tok/s | chat, coding, agents, long documents |
| LFM2.5 2.6B Liquid AI | Q8_0 | fits | 3.9 GB | 124k128k with q8 cache | 93tok/sest. | 2049 tok/s | chat, coding, agents, long documents |
| LFM2.5 VL 3B Liquid AI | Q8_0 | fits | 3.9 GB | 32k | 93tok/sest. | 1808 tok/s | chat, coding, agents, long documents |
| Spark X2.5 4B XHToken | mlx-4bit | fits | 3.9 GB | 190k319k with q8 cache | 92tok/sest. | 1449 tok/s | chat, coding, agents, long documents |
| Kimi VL A3B Thinking 2506 Moonshot AI | Q4_K_M | fits | 10.8 GB | 24k40k with q8 cache | 86tok/sest. | 1899 tok/s | chat, coding |
| Mellum2.1 12B A2.5B Thinking JetBrains | Q5_K_M | fits | 10.2 GB | 70k106k with q8 cache | 85tok/sest. | 2253 tok/s | chat, coding, agents, long documents |
| granite vision 4.1 4b IBM | Q8_0 | fits | 4.7 GB | 87k128k with q8 cache | 73tok/sest. | 1356 tok/s | chat, coding |
| Ministral 3 3B Instruct 2512 Mistral AI | Q8_0 | fits | 5.0 GB | 67k122k with q8 cache | 69tok/sest. | 1389 tok/s | chat, coding, agents |
| granite 4.2 3b IBM | Q8_0 | fits | 5.0 GB | 84k128k with q8 cache | 69tok/sest. | 1458 tok/s | chat, coding, agents |
| FrogNano 4B 2609 Microsoft | Q8_0 | fits | 5.3 GB | 171k256k with q8 cache | 63tok/sest. | 1331 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Qwen 7B DeepSeek | Q8_0 | fits | 8.7 GB | 52k90k with q8 cache | 37tok/sest. | 777 tok/s | chat, coding |
| GLM 4 9B 0414 Z.ai | mlx-5bit | fits | 8.7 GB | 32k | 36tok/sest. | 611 tok/s | chat, coding |
| DeepSeek R1 Distill Llama 8B DeepSeek | Q8_0 | fits | 9.7 GB | 20k37k with q8 cache | 33tok/sest. | 720 tok/s | chat |
| Ministral 3 8B Reasoning 2512 Mistral AI | Q8_0 | fits | 10.2 GB | 16k29k with q8 cache | 31tok/sest. | 652 tok/s | chat |
| GLM 4.6V Flash Z.ai | Q8_0 | fits | 10.4 GB | 30k52k with q8 cache | 30tok/sest. | 564 tok/s | chat, coding |
| DeepSeek R1 Distill Qwen 14B DeepSeek | Q4_K_M | fits | 10.6 GB | 12k22k with q8 cache | 30tok/sest. | 391 tok/s | chat |
| Ministral 3 14B Reasoning 2512 Mistral AI | Q5_K_M | fits | 11.0 GB | 10k19k with q8 cache | 29tok/sest. | 429 tok/s | chat |
| Qwen 3.8 27B Alibaba Qwen | UD-Q4_K_M | offload | 16.7 GB | – | 7tok/sest. | 184 tok/s | – |
| Qwen 3.6 27B Alibaba Qwen | Q4_K_M | offload | 17.0 GB | – | 7tok/sest. | 182 tok/s | – |
| Gemma 4 31B | Q4_K_M | offload | 19.2 GB | – | 5tok/sest. | 149 tok/s | – |
| Qwen 3.6 35B-A3B Alibaba Qwen | MXFP4 | offload | 21.2 GB | – | 28tok/sest. | 1402 tok/s | chat, coding |
| Qwen 3.5 27B Alibaba Qwen | Q4_K_M | offload | 17.0 GB | – | 7tok/sest. | 182 tok/s | – |
| Gemma 4 26B-A4B | MXFP4 | offload | 16.5 GB | – | 25tok/sest. | 1207 tok/s | chat, coding |
| Qwen 3.5 35B-A3B Alibaba Qwen | Q4_K_M | offload | 21.5 GB | – | 27tok/sest. | 1393 tok/s | chat, coding |
| GLM 4.7 Flash Z.ai | Q4_K_M | offload | 18.2 GB | – | 19tok/sest. | 843 tok/s | chat |
| Devstral Small 2 24B Mistral AI | Q4_K_M | offload | 15.3 GB | – | 9tok/sest. | 217 tok/s | – |
| Nemotron 3.5 Lightning 30B-A3B NVIDIA | MXFP4 | offload | 22.5 GB | – | 24tok/sest. | 1423 tok/s | chat, coding, long documents |
| gpt-oss 20B OpenAI | Q4_K_M | offload | 11.8 GB | – | 64tok/sest. | 1395 tok/s | chat, coding, agents |
| Kimi Linear 48B-A3B Moonshot AI | Q4_K_M | offload | 28.8 GB | – | 24tok/sest. | 1173 tok/s | chat, coding, long documents |
| Qwen3 Coder 30B-A3B Alibaba Qwen | Q4_K_M | offload | 18.8 GB | – | 24tok/sest. | 1073 tok/s | chat |
| Qwen3 32B Alibaba Qwen | Q4_K_M | offload | 21.2 GB | – | 4tok/sest. | 126 tok/s | – |
| Mistral Small 3.2 24B Mistral AI | Q4_K_M | offload | 15.3 GB | – | 9tok/sest. | 217 tok/s | – |
| Gemma 3 27B | Q4_K_M | offload | 17.2 GB | – | 7tok/sest. | 183 tok/s | – |
| DeepSeek R1 Distill Qwen 32B DeepSeek | Q4_K_M | offload | 21.2 GB | – | 4tok/sest. | 132 tok/s | – |
| Kimi VL A3B Moonshot AI | mlx-5bit | offload | 12.0 GB | – | 62tok/sest. | 1838 tok/s | chat, coding, agents |
| LFM2.5 24B A2B Liquid AI | Q4_K_M | offload | 15.7 GB | – | 40tok/sest. | 2034 tok/s | chat |
| GLM 4.1V 9B Thinking Z.ai | mlx-8bit | offload | 12.0 GB | – | 20tok/sest. | 545 tok/s | chat |
| Devstral Small 2507 Mistral AI | Q4_K_M | offload | 15.3 GB | – | 9tok/sest. | 221 tok/s | – |
| Magistral Small 2506 Mistral AI | Q4_K_M | offload | 15.3 GB | – | 9tok/sest. | 221 tok/s | – |
| Magistral Small 2507 Mistral AI | Q4_K_M | offload | 15.3 GB | – | 9tok/sest. | 221 tok/s | – |
| Magistral Small 2509 Mistral AI | Q4_K_M | offload | 15.3 GB | – | 9tok/sest. | 217 tok/s | – |
| Muse Glimmer 30B meta-models | Q4_K_M | offload | 17.0 GB | – | 7tok/sest. | 168 tok/s | – |
| granite 4.2 30b IBM | Q4_K_M | offload | 19.5 GB | – | 5tok/sest. | 155 tok/s | – |
| Qwen3.8 27B OBLITERATED OBLITERATUS | mlx-4bit | offload | 20.6 GB | – | 5tok/sest. | 166 tok/s | – |
| DeepSeek V4 Flash DeepSeek | UD-Q2_K_XL | won't run | 91.6 GB | – | – | – | – |
| GLM 5.3 Flash Z.ai | UD-Q2_K_XL | won't run | 102.2 GB | – | – | – | – |
| Qwen 3.8 Flash Next Alibaba Qwen | UD-Q2_K_XL | won't run | 74.5 GB | – | – | – | – |
| MiniMax M2.7 MiniMax | UD-Q2_K_XL | won't run | 72.8 GB | – | – | – | – |
| Qwen 3.5 122B-A10B Alibaba Qwen | UD-Q2_K_XL | won't run | 40.0 GB | – | – | – | – |
| Nemotron 3 Super 120B-A12B NVIDIA | UD-Q2_K_XL | won't run | 51.9 GB | – | – | – | – |
| Qwen3 Coder Next Alibaba Qwen | Q4_K_M | won't run | 46.2 GB | – | – | – | – |
| gpt-oss 120B OpenAI | Q4_K_M | won't run | 59.5 GB | – | – | – | – |
| Mistral Small 4 119B Mistral AI | UD-Q2_K_XL | won't run | 38.4 GB | – | – | – | – |
| GLM 4.5 Air Z.ai | Q2_K | won't run | 44.4 GB | – | – | – | – |
| Qwen3 Next 80B-A3B Alibaba Qwen | Q4_K_M | won't run | 46.2 GB | – | – | – | – |
| Llama 3.3 70B Meta | Q4_K_M | won't run | 42.8 GB | – | – | – | – |
| DeepSeek R1 DeepSeek | Q4_K_M | won't run | 377.9 GB | – | – | – | – |
| DeepSeek R1 0528 DeepSeek | Q4_K_M | won't run | 378.4 GB | – | – | – | – |
| DeepSeek R1 Distill Llama 70B DeepSeek | Q4_K_M | won't run | 42.8 GB | – | – | – | – |
| DeepSeek R1 Zero DeepSeek | Q4_K_M | won't run | 377.9 GB | – | – | – | – |
| DeepSeek V3 DeepSeek | Q4_K_M | won't run | 377.9 GB | – | – | – | – |
| DeepSeek V3 0324 DeepSeek | Q4_K_M | won't run | 377.9 GB | – | – | – | – |
| DeepSeek V3.1 DeepSeek | Q4_K_M | won't run | 378.8 GB | – | – | – | – |
| DeepSeek V3.1 Terminus DeepSeek | Q4_K_M | won't run | 378.8 GB | – | – | – | – |
| DeepSeek V3.2 DeepSeek | Q4_K_M | won't run | 378.8 GB | – | – | – | – |
| DeepSeek V4 Flash 0731 DeepSeek | MXFP4 | won't run | 147.1 GB | – | – | – | – |
| DeepSeek V4 Flash Vision Exp DeepSeek | MXFP4 | won't run | 155.8 GB | – | – | – | – |
| Devstral 2 123B Instruct 2512 Mistral AI | Q4_K_M | won't run | 73.3 GB | – | – | – | – |
| GLM 4.5 Z.ai | Q4_K_M | won't run | 205.2 GB | – | – | – | – |
| GLM 4.5V Z.ai | Q4_K_M | won't run | 61.4 GB | – | – | – | – |
| GLM 4.6 Z.ai | Q4_K_M | won't run | 204.4 GB | – | – | – | – |
| GLM 4.6V Z.ai | Q4_K_M | won't run | 67.8 GB | – | – | – | – |
| GLM 4.7 Z.ai | Q4_K_M | won't run | 205.2 GB | – | – | – | – |
| GLM 5 Z.ai | MXFP4 | won't run | 383.8 GB | – | – | – | – |
| GLM 5.1 Z.ai | MXFP4 | won't run | 421.4 GB | – | – | – | – |
| GLM 5.2 Z.ai | UD-Q4_K_M | won't run | 435.3 GB | – | – | – | – |
| GLM 5.3 Z.ai | Q8_0 | won't run | 747.8 GB | – | – | – | – |
| Kimi Dev 72B Moonshot AI | Q6_K | won't run | 63.2 GB | – | – | – | – |
| Kimi K2 Moonshot AI | Q4_K_M | won't run | 579.4 GB | – | – | – | – |
| Kimi K2 Instruct 0905 Moonshot AI | Q4_K_M | won't run | 579.9 GB | – | – | – | – |
| Kimi K2 Thinking Moonshot AI | Q4_K_M | won't run | 579.9 GB | – | – | – | – |
| Kimi K2.5 Moonshot AI | Q4_K_M | won't run | 579.9 GB | – | – | – | – |
| MiniMax M2 MiniMax | Q4_K_M | won't run | 131.5 GB | – | – | – | – |
| MiniMax M2.1 MiniMax | Q4_K_M | won't run | 131.5 GB | – | – | – | – |
| MiniMax M2.5 MiniMax | MXFP4 | won't run | 118.0 GB | – | – | – | – |
| MiniMax M3 MiniMax | MXFP4 | won't run | 241.4 GB | – | – | – | – |
| Mistral Medium 3.5 128B Mistral AI | Q4_K_M | won't run | 73.3 GB | – | – | – | – |
Specifications
From the manufacturer's published specifications; sources are linked where we have them.
- GPU
- GeForce RTX 4070
- Memory
- 12 GB GDDR6X
- Memory bus
- 192-bit
- Memory bandwidth
- 504.2 GB/s
- FP16 compute
- 29.146 TFLOPS
- Software
- CUDA
- Board power
- 200 W
- PCIe generation
- Gen 4
- PCIe lanes
- x16
Compare with
Questions
- How much memory does the NVIDIA GeForce RTX 4070 have for AI models?
- The NVIDIA GeForce RTX 4070 has 12 GB of GDDR6X memory with 504.2 GB/s of bandwidth, which sets how fast it generates text.
- What is the best model the NVIDIA GeForce RTX 4070 can run locally?
- Gemma 4 12B (Q6_K) fits fully in its memory at 8k context and generates at about 31 tok/s, an estimate from its memory bandwidth. 34 of the 125 models we track fit at that context.
- Can the NVIDIA GeForce RTX 4070 run DeepSeek V4 Flash?
- No: even at UD-Q2_K_XL, DeepSeek V4 Flash needs about 92 GB at 8k context, more than the NVIDIA GeForce RTX 4070 and system RAM can hold.
- How much does the NVIDIA GeForce RTX 4070 cost in the US?
- We have no current price for the NVIDIA GeForce RTX 4070 in the US.
Speeds on this page are for 8k context. See the best hardware for local AI at every budget.