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 32k context it runs 56 of the 125 open models we track fully in memory; the most capable is Qwen 3.8 27B at UD-Q5_K_M, at about 14 tok/s (estimated). We have no current price for the NVIDIA GeForce RTX 4070 in the UK.
Updated . Speeds are estimates; how they are worked out.
Prices
Every retailer we track in the UK, cheapest first. Buy links may earn us a commission.
No retailer in this market lists it right now.
What two of them run
Every model in the catalogue, with the best quant for this hardware. Estimates, not measurements.
| Model | Quant | Fits | Memory | Max context | Decode @32k | Prefill | Good for |
|---|---|---|---|---|---|---|---|
| Qwen 3.8 27B Alibaba Qwen | UD-Q5_K_M | fits | 21.8 GB | 46k80k with q8 cache | 14tok/sest. | 212 tok/s | chat |
| Qwen 3.6 27B Alibaba Qwen | Q5_K_M | fits | 21.6 GB | 49k86k with q8 cache | 14tok/sest. | 212 tok/s | chat |
| Gemma 4 31B | Q4_K_M | fits | 21.6 GB | 45k89k with q8 cache | 14tok/sest. | 189 tok/s | chat |
| Qwen 3.6 35B-A3B Alibaba Qwen | Q4_K_M | fits | 22.7 GB | 34k54k with q8 cache | 72tok/sest. | 1510 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 27B Alibaba Qwen | Q5_K_M | fits | 21.7 GB | 48k84k with q8 cache | 14tok/sest. | 212 tok/s | chat |
| Gemma 4 26B-A4B | UD-Q5_K_M | fits | 21.8 GB | 71k116k with q8 cache | 45tok/sest. | 1344 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 35B-A3B Alibaba Qwen | Q4_K_M | fits | 22.5 GB | 44k70k with q8 cache | 73tok/sest. | 1510 tok/s | chat, coding, agents, long documents |
| GLM 4.7 Flash Z.ai | Q4_K_M | fits | 20.0 GB | 80k139k with q8 cache | 39tok/sest. | 789 tok/s | chat, coding, agents |
| Devstral Small 2 24B Mistral AI | Q5_K_M | fits | 21.9 GB | 37k68k with q8 cache | 14tok/sest. | 223 tok/s | chat |
| gpt-oss 20B OpenAI | MXFP4 | fits | 13.3 GB | 128k | 58tok/sest. | 1128 tok/s | chat, coding, agents, long documents |
| Qwen3 Coder 30B-A3B Alibaba Qwen | Q4_K_M | fits | 21.6 GB | 44k79k with q8 cache | 37tok/sest. | 695 tok/s | chat, coding, agents |
| Gemma 4 12B | Q8_0 | fits | 13.9 GB | 256k | 23tok/sest. | 491 tok/s | chat, coding, long documents |
| Qwen 3.5 9B Alibaba Qwen | Q8_0 | fits | 11.2 GB | 256k | 29tok/sest. | 611 tok/s | chat, coding, long documents |
| Mistral Small 3.2 24B Mistral AI | Q5_K_M | fits | 21.9 GB | 37k68k with q8 cache | 14tok/sest. | 223 tok/s | chat |
| Gemma 3 27B | Q5_K_M | fits | 22.1 GB | 39k74k with q8 cache | 14tok/sest. | 225 tok/s | chat |
| Ministral 3 14B Mistral AI | Q8_0 | fits | 19.7 GB | 51k93k with q8 cache | 16tok/sest. | 340 tok/s | chat |
| Qwen3 14B Alibaba Qwen | Q8_0 | fits | 20.9 GB | 40k | 15tok/sest. | 305 tok/s | chat |
| DeepSeek R1 0528 Qwen3 8B DeepSeek | Q8_0 | fits | 13.9 GB | 92k128k with q8 cache | 23tok/sest. | 504 tok/s | chat, coding |
| Qwen3 8B Alibaba Qwen | Q8_0 | fits | 13.9 GB | 40k | 23tok/sest. | 504 tok/s | chat, coding |
| Phi-4 14B Microsoft | Q8_0 | fits | 22.0 GB | 16k | 14tok/sest. | 307 tok/s | chat |
| Qwen 3.5 4B Alibaba Qwen | Q8_0 | fits | 6.5 GB | 256k | 56tok/sest. | 1144 tok/s | chat, coding, long documents |
| Ministral 3 8B Mistral AI | Q8_0 | fits | 13.9 GB | 95k173k with q8 cache | 23tok/sest. | 486 tok/s | chat, coding |
| Granite 4.2 8B IBM | Q8_0 | fits | 15.0 GB | 80k128k with q8 cache | 21tok/sest. | 463 tok/s | chat, coding |
| Gemma 4 E4B | Q8_0 | fits | 9.8 GB | 128k | 34tok/sest. | 768 tok/s | chat, coding, long documents |
| LFM2.5 8B-A1B Liquid AI | Q8_0 | fits | 11.2 GB | 125k | 70tok/sest. | 2512 tok/s | chat, coding, long documents |
| Llama 3.1 8B Meta | Q6_K | fits | 11.4 GB | 118k128k with q8 cache | 29tok/sest. | 532 tok/s | chat, coding |
| Qwen2.5 0.5B Alibaba Qwen | Q8_0 | fits | 2.2 GB | 32k | 335tok/sest. | 5471 tok/s | chat, coding, long documents |
| Qwen3.5 0.8B Alibaba Qwen | Q8_0 | fits | 2.4 GB | 256k | 254tok/sest. | 5139 tok/s | chat, coding, agents, long documents |
| Qwen3.5 2B Alibaba Qwen | Q5_K_M | fits | 3.0 GB | 256k | 168tok/sest. | 2450 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Qwen 1.5B DeepSeek | Q8_0 | fits | 3.9 GB | 128k | 110tok/sest. | 2058 tok/s | chat, coding, long documents |
| MiniCPM5 2B openbmb | Q8_0 | fits | 5.1 GB | 128k | 77tok/sest. | 1229 tok/s | chat, coding, agents, long documents |
| Spark X2.5 4B XHToken | mlx-4bit | fits | 5.3 GB | 458k767k with q8 cache | 73tok/sest. | 1207 tok/s | chat, coding, agents, long documents |
| LFM2.5 2.6B Liquid AI | Q8_0 | fits | 5.8 GB | 128k | 64tok/sest. | 1392 tok/s | chat, coding, agents, long documents |
| LFM2.5 VL 3B Liquid AI | Q8_0 | fits | 5.8 GB | 32k | 64tok/sest. | 1277 tok/s | chat, coding, agents, long documents |
| Mellum2.1 12B A2.5B Thinking JetBrains | Q8_0 | fits | 13.8 GB | 128k | 59tok/sest. | 1814 tok/s | chat, coding, agents, long documents |
| FrogNano 4B 2609 Microsoft | Q8_0 | fits | 6.6 GB | 256k | 54tok/sest. | 1144 tok/s | chat, coding, agents, long documents |
| granite vision 4.1 4b IBM | Q8_0 | fits | 7.2 GB | 128k | 50tok/sest. | 892 tok/s | chat, coding, long documents |
| granite 4.2 3b IBM | Q8_0 | fits | 7.4 GB | 128k | 48tok/sest. | 935 tok/s | chat, coding, agents, long documents |
| Kimi VL A3B Thinking 2506 Moonshot AI | Q8_0 | fits | 18.0 GB | 128k | 46tok/sest. | 1363 tok/s | chat, coding, long documents |
| LFM2.5 24B A2B Liquid AI | Q5_K_M | fits | 21.0 GB | 53k94k with q8 cache | 44tok/sest. | 1400 tok/s | chat, coding, agents |
| Kimi VL A3B Moonshot AI | mlx-8bit | fits | 18.8 GB | 128k | 44tok/sest. | 1363 tok/s | chat, coding, agents, long documents |
| Ministral 3 3B Instruct 2512 Mistral AI | Q8_0 | fits | 7.9 GB | 170k256k with q8 cache | 44tok/sest. | 894 tok/s | chat, coding, agents, long documents |
| GLM 4 9B 0414 Z.ai | mlx-5bit | fits | 10.2 GB | 32k | 33tok/sest. | 444 tok/s | chat, coding |
| DeepSeek R1 Distill Qwen 7B DeepSeek | Q8_0 | fits | 10.6 GB | 128k | 31tok/sest. | 601 tok/s | chat, coding, long documents |
| GLM 4.6V Flash Z.ai | Q8_0 | fits | 11.9 GB | 128k | 28tok/sest. | 419 tok/s | chat, coding |
| DeepSeek R1 Distill Llama 8B DeepSeek | Q8_0 | fits | 13.2 GB | 105k128k with q8 cache | 24tok/sest. | 532 tok/s | chat, coding |
| GLM 4.1V 9B Thinking Z.ai | mlx-8bit | fits | 13.5 GB | 64k | 24tok/sest. | 419 tok/s | chat, coding |
| Ministral 3 8B Reasoning 2512 Mistral AI | Q8_0 | fits | 13.9 GB | 95k173k with q8 cache | 23tok/sest. | 486 tok/s | chat, coding |
| Ministral 3 14B Reasoning 2512 Mistral AI | Q8_0 | fits | 19.7 GB | 51k93k with q8 cache | 16tok/sest. | 340 tok/s | chat |
| Muse Glimmer 30B meta-models | Q5_K_M | fits | 20.5 GB | 128k | 15tok/sest. | 204 tok/s | chat |
| DeepSeek R1 Distill Qwen 14B DeepSeek | Q8_0 | fits | 21.9 GB | 36k67k with q8 cache | 14tok/sest. | 287 tok/s | chat |
| Devstral Small 2507 Mistral AI | Q5_K_M | fits | 21.9 GB | 37k68k with q8 cache | 14tok/sest. | 227 tok/s | chat |
| Magistral Small 2506 Mistral AI | Q5_K_M | fits | 21.9 GB | 37k40k with q8 cache | 14tok/sest. | 227 tok/s | chat |
| Magistral Small 2507 Mistral AI | Q5_K_M | fits | 21.9 GB | 37k40k with q8 cache | 14tok/sest. | 227 tok/s | chat |
| Magistral Small 2509 Mistral AI | Q5_K_M | fits | 21.9 GB | 37k68k with q8 cache | 14tok/sest. | 223 tok/s | chat |
| Qwen3.8 27B OBLITERATED OBLITERATUS | mlx-4bit | fits | 22.7 GB | 33k59k with q8 cache | 14tok/sest. | 212 tok/s | – |
| Qwen 3.5 122B-A10B Alibaba Qwen | UD-Q2_K_XL | offload | 41.1 GB | – | 14tok/sest. | 359 tok/s | chat |
| Qwen3 Coder Next Alibaba Qwen | Q4_K_M | offload | 47.3 GB | – | 16tok/sest. | 814 tok/s | chat |
| Mistral Small 4 119B Mistral AI | UD-Q2_K_XL | offload | 39.4 GB | – | 19tok/sest. | 407 tok/s | chat |
| Qwen3 Next 80B-A3B Alibaba Qwen | Q4_K_M | offload | 47.3 GB | – | 20tok/sest. | 968 tok/s | chat |
| Nemotron 3.5 Lightning 30B-A3B NVIDIA | MXFP4 | offload | 23.1 GB | 3k4k with q8 cache | 71tok/sest. | 1707 tok/s | chat, coding, agents, long documents |
| Kimi Linear 48B-A3B Moonshot AI | Q4_K_M | offload | 29.6 GB | – | 41tok/sest. | 1115 tok/s | chat, coding, long documents |
| Qwen3 32B Alibaba Qwen | Q4_K_M | offload | 27.7 GB | 12k23k with q8 cache | 6tok/sest. | 117 tok/s | – |
| DeepSeek R1 Distill Qwen 32B DeepSeek | Q4_K_M | offload | 27.8 GB | 12k23k with q8 cache | 6tok/sest. | 134 tok/s | – |
| granite 4.2 30b IBM | Q4_K_M | offload | 26.1 GB | 19k35k with q8 cache | 7tok/sest. | 159 tok/s | chat |
| DeepSeek V4 Flash DeepSeek | UD-Q2_K_XL | won't run | 94.2 GB | – | – | – | – |
| GLM 5.3 Flash Z.ai | UD-Q2_K_XL | won't run | 103.0 GB | – | – | – | – |
| Qwen 3.8 Flash Next Alibaba Qwen | UD-Q2_K_XL | won't run | 75.6 GB | – | – | – | – |
| MiniMax M2.7 MiniMax | UD-Q2_K_XL | won't run | 79.2 GB | – | – | – | – |
| Nemotron 3 Super 120B-A12B NVIDIA | UD-Q2_K_XL | won't run | 52.6 GB | – | – | – | – |
| gpt-oss 120B OpenAI | Q4_K_M | won't run | 60.9 GB | – | – | – | – |
| GLM 4.5 Air Z.ai | Q2_K | won't run | 49.2 GB | – | – | – | – |
| Llama 3.3 70B Meta | Q4_K_M | won't run | 50.9 GB | – | – | – | – |
| DeepSeek R1 DeepSeek | Q4_K_M | won't run | 380.1 GB | – | – | – | – |
| DeepSeek R1 0528 DeepSeek | Q4_K_M | won't run | 380.6 GB | – | – | – | – |
| DeepSeek R1 Distill Llama 70B DeepSeek | Q4_K_M | won't run | 50.9 GB | – | – | – | – |
| DeepSeek R1 Zero DeepSeek | Q4_K_M | won't run | 380.1 GB | – | – | – | – |
| DeepSeek V3 DeepSeek | Q4_K_M | won't run | 380.1 GB | – | – | – | – |
| DeepSeek V3 0324 DeepSeek | Q4_K_M | won't run | 380.1 GB | – | – | – | – |
| DeepSeek V3.1 DeepSeek | Q4_K_M | won't run | 381.0 GB | – | – | – | – |
| DeepSeek V3.1 Terminus DeepSeek | Q4_K_M | won't run | 381.0 GB | – | – | – | – |
| DeepSeek V3.2 DeepSeek | Q4_K_M | won't run | 381.0 GB | – | – | – | – |
| DeepSeek V4 Flash 0731 DeepSeek | MXFP4 | won't run | 149.6 GB | – | – | – | – |
| DeepSeek V4 Flash Vision Exp DeepSeek | MXFP4 | won't run | 158.4 GB | – | – | – | – |
| Devstral 2 123B Instruct 2512 Mistral AI | Q4_K_M | won't run | 82.0 GB | – | – | – | – |
| GLM 4.5 Z.ai | Q4_K_M | won't run | 214.4 GB | – | – | – | – |
| GLM 4.5V Z.ai | Q4_K_M | won't run | 66.3 GB | – | – | – | – |
| GLM 4.6 Z.ai | Q4_K_M | won't run | 213.6 GB | – | – | – | – |
| GLM 4.6V Z.ai | Q4_K_M | won't run | 72.6 GB | – | – | – | – |
| GLM 4.7 Z.ai | Q4_K_M | won't run | 214.4 GB | – | – | – | – |
| GLM 5 Z.ai | MXFP4 | won't run | 386.4 GB | – | – | – | – |
| GLM 5.1 Z.ai | MXFP4 | won't run | 424.0 GB | – | – | – | – |
| GLM 5.2 Z.ai | UD-Q4_K_M | won't run | 437.9 GB | – | – | – | – |
| GLM 5.3 Z.ai | Q8_0 | won't run | 750.4 GB | – | – | – | – |
| Kimi Dev 72B Moonshot AI | Q6_K | won't run | 71.2 GB | – | – | – | – |
| Kimi K2 Moonshot AI | Q4_K_M | won't run | 581.6 GB | – | – | – | – |
| Kimi K2 Instruct 0905 Moonshot AI | Q4_K_M | won't run | 582.0 GB | – | – | – | – |
| Kimi K2 Thinking Moonshot AI | Q4_K_M | won't run | 582.0 GB | – | – | – | – |
| Kimi K2.5 Moonshot AI | Q4_K_M | won't run | 582.0 GB | – | – | – | – |
| MiniMax M2 MiniMax | Q4_K_M | won't run | 137.9 GB | – | – | – | – |
| MiniMax M2.1 MiniMax | Q4_K_M | won't run | 137.9 GB | – | – | – | – |
| MiniMax M2.5 MiniMax | MXFP4 | won't run | 124.3 GB | – | – | – | – |
| MiniMax M3 MiniMax | MXFP4 | won't run | 244.7 GB | – | – | – | – |
| Mistral Medium 3.5 128B Mistral AI | Q4_K_M | won't run | 82.0 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?
- Qwen 3.8 27B (UD-Q5_K_M) fits fully in its memory at 32k context and generates at about 14 tok/s, an estimate from its memory bandwidth. 56 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 94 GB at 32k context, more than the NVIDIA GeForce RTX 4070 and system RAM can hold.
- How much does the NVIDIA GeForce RTX 4070 cost in the UK?
- We have no current price for the NVIDIA GeForce RTX 4070 in the UK.
Speeds on this page are for 32k context. See the best hardware for local AI at every budget.