GPUs · AMD
Can the RX 6700 XT run local LLMs?
With 12 GB of VRAM (about 10.8 GB usable) and 384 GB/s of memory bandwidth, the RX 6700 XT runs 16 of 39 popular open models fully on the GPU at 8K context. The largest is Mistral Nemo 12B at Q4_K_M, at roughly 35 tokens per second.
Best picks for the RX 6700 XT
Every model on the RX 6700 XT
Results at 8K context, picking the highest quality quantization that fits. "Offload" means part of the model runs from system RAM (we assume 32 GB), which works but is much slower.
| Model | Status | Quant | Memory | Speed |
|---|---|---|---|---|
| Mistral Nemo 12B | Runs well | Q4_K_M | 9.2 GB | ~35 tok/s |
| Gemma 4 12B | Runs well | Q4_K_M | 11 GB | ~35 tok/s |
| Qwen3.5 9B | Runs well | Q6_K | 9.4 GB | ~32 tok/s |
| Qwen3 8B | Runs well | Q8_0 | 11 GB | ~29 tok/s |
| Llama 3.1 8B | Runs well | Q8_0 | 10 GB | ~29 tok/s |
| DeepSeek-R1 Distill 8B | Runs well | Q8_0 | 10 GB | ~29 tok/s |
| Qwen2.5-Coder 7B | Runs well | Q8_0 | 9.8 GB | ~31 tok/s |
| Mistral 7B v0.3 | Runs well | Q8_0 | 9.4 GB | ~33 tok/s |
| Qwen3.5 4B | Runs well | Q8_0 | 6.4 GB | ~50 tok/s |
| Gemma 3 4B | Runs well | Q8_0 | 6.9 GB | ~55 tok/s |
| Qwen3 4B | Runs well | Q8_0 | 5.8 GB | ~59 tok/s |
| Phi-4 Mini 3.8B | Runs well | Q8_0 | 5.2 GB | ~62 tok/s |
| Llama 3.2 3B | Runs well | Q8_0 | 4.6 GB | ~74 tok/s |
| Llama 3.2 1B | Runs well | Q8_0 | 1.6 GB | ~200 tok/s |
| Gemma 3 1B | Runs well | Q8_0 | 1.4 GB | ~240 tok/s |
| Qwen3 0.6B | Runs well | Q8_0 | 0.9 GB | ~390 tok/s |
| Phi-4 14B | Offload | Q4_K_M | 11 GB | ~4.3 tok/s |
| Gemma 3 12B | Offload | Q4_K_M | 11 GB | ~5.2 tok/s |
| Qwen3 14B | Offload | Q4_K_M | 11 GB | ~4.3 tok/s |
| DeepSeek-R1 Distill 14B | Offload | Q4_K_M | 11 GB | ~4.3 tok/s |
| gpt-oss 20B | Offload | Q4_K_M | 14 GB | ~9.0 tok/s |
| Codestral 22B | Offload | Q4_K_M | 16 GB | ~2.8 tok/s |
| Devstral Small 24B | Offload | Q4_K_M | 17 GB | ~2.6 tok/s |
| Mistral Small 3 24B | Offload | Q4_K_M | 18 GB | ~2.6 tok/s |
| Gemma 4 26B-A4B (MoE) | Offload | Q4_K_M | 19 GB | ~8.9 tok/s |
| Qwen3.6 27B | Offload | Q4_K_M | 19 GB | ~2.2 tok/s |
| Qwen3 30B-A3B (MoE) | Offload | Q4_K_M | 21 GB | ~10 tok/s |
| Qwen3-Coder 30B-A3B | Offload | Q4_K_M | 21 GB | ~10 tok/s |
| Gemma 3 27B | Offload | Q4_K_M | 22 GB | ~2.2 tok/s |
| Qwen3 32B | Offload | Q4_K_M | 23 GB | ~1.9 tok/s |
| DeepSeek-R1 Distill 32B | Offload | Q4_K_M | 23 GB | ~1.9 tok/s |
| Qwen2.5-Coder 32B | Offload | Q4_K_M | 23 GB | ~1.9 tok/s |
| Qwen3.6 35B-A3B (MoE) | Offload | Q4_K_M | 23 GB | ~11 tok/s |
| Gemma 4 31B | Offload | Q4_K_M | 25 GB | ~2.0 tok/s |
| Llama 3.3 70B | Won’t fit | — | 48 GB | — |
| DeepSeek-R1 Distill 70B | Won’t fit | — | 48 GB | — |
| Llama 4 Scout 109B-A17B | Won’t fit | — | 73 GB | — |
| gpt-oss 120B | Won’t fit | — | 77 GB | — |
| Qwen3 235B-A22B (MoE) | Won’t fit | — | 155 GB | — |
What this means in practice
The RX 6700 XT is comfortable with models up to about 12B to 14B parameters. gpt-oss 20B is the most useful step up beyond that: it needs about 14 GB, so part of it runs from system RAM, but as a mixture-of-experts model it stays reasonably quick at around 9.0 tokens per second. Models from Llama 3.3 70B upward need 48 GB or more and are out of reach. AMD cards work with Ollama and LM Studio through ROCm or Vulkan; setup is smoother on Linux and recent Windows drivers.
The next memory tier up is 16 GB: RTX 5080, RTX 5070 Ti, RTX 4080 / 4080 Super. Other 12 GB cards: RTX 5070, RTX 4070 / 4070 Super, RTX 3060, RTX 4080 Laptop, RX 7700 XT, Arc B580.
Compare the RX 6700 XT
Questions
What is the biggest model the RX 6700 XT can run?
At 8K context the largest model that fits entirely in its 12 GB of VRAM is Mistral Nemo 12B at Q4_K_M, at roughly 35 tokens per second.
Can the RX 6700 XT run 70B models like Llama 3.3 70B?
No. It needs about 48 GB at Q4, which is far beyond 12 GB even with system RAM offload.
What is the best model for coding on the RX 6700 XT?
Qwen3.5 9B at Q6_K fits fully and runs at about 32 tokens per second. Pair it with an editor extension like Continue pointed at Ollama.
Estimates based on memory size and bandwidth; real speed varies with drivers, software and context length. See how we calculate VRAM.