- Services
Every open model in our directory, ranked by how well it runs on AMD Radeon RX 7700 XT (12GB) at 4-bit. Grades weigh whether the model fits in memory and how fast it should run.
AMD Radeon RX 7700 XT has 12 GB of VRAM, enough to run 204 of the 301 open models we track at 4-bit. The largest that still fits well is AliceAI-Foundation-80B-A3B-Base (81.3B), and the top-rated is AliceAI-Foundation-80B-A3B-Base at about 41 tok/s.
204 of 301 models run comfortably on AMD Radeon RX 7700 XT.
See full AMD Radeon RX 7700 XT specs and pricingThe top models this device can run at 4-bit, ranked by fit and speed.
| Model | Grade | Speed | VRAM |
|---|---|---|---|
| AliceAI-Foundation-80B-A3B-Baseyandex | SS | 40.8 tok/s | 8.5 GB |
| North Mini CodeCohere | SS | 41.5 tok/s | 8.4 GB |
| Nemotron 3 Nano OmniNVIDIA | SS | 40.8 tok/s | 8.5 GB |
| Qwen3.6 35B-A3BAlibaba | SS | 40.8 tok/s | 8.5 GB |
| Qwen3.5-35B-A3BAlibaba | SS | 40.8 tok/s | 8.5 GB |
| MiniCPM-V 4.7 35B-A3Bopenbmb | SS | 40.8 tok/s | 8.5 GB |
| Qwen3-30B-A3BAlibaba | SS | 64.6 tok/s | 5.4 GB |
| Llama 2 13B ChatMeta | SS | 41.1 tok/s | 8.5 GB |
| LensVLM-9BApple | SS | 57.8 tok/s | 6.0 GB |
| Carnice-9b for Hermes agentkai-os | SS | 57.8 tok/s | 6.0 GB |
| Llama 3 8B InstructMeta | SS | 61.4 tok/s | 5.7 GB |
| Apertus 8BEPFL, ETH Zurich, CSCS | SS | 64.4 tok/s | 5.4 GB |
| 81.29B(3B active) | SS | 40.8 tok/s | 8.5 GB | ||
North Mini CodeCohere | 30B(3B active) | SS | 41.5 tok/s | 8.4 GB | |
Nemotron 3 Nano OmniNVIDIA | 30B(3B active) | SS | 40.8 tok/s | 8.5 GB | |
Qwen3.6 35B-A3BAlibaba | 35B(3B active) | SS | 40.8 tok/s | 8.5 GB | |
Qwen3.5-35B-A3BAlibaba | 35B(3B active) | SS | 40.8 tok/s | 8.5 GB | |
MiniCPM-V 4.7 35B-A3Bopenbmb | 35.21B(3B active) | SS | 40.8 tok/s | 8.5 GB | |
Qwen3-30B-A3BAlibaba | 30B(3B active) | SS | 64.6 tok/s | 5.4 GB | |
Llama 2 13B ChatMeta | 13B | SS | 41.1 tok/s | 8.5 GB | |
Ad Tools for AI BuildersAll the hardware and tools you need to run private AI at your office or at home.Shop Now | |||||
LensVLM-9BApple | 9B | SS | 57.8 tok/s | 6.0 GB | |
| 9B | SS | 57.8 tok/s | 6.0 GB | ||
| 8B | SS | 61.4 tok/s | 5.7 GB | ||
Apertus 8BEPFL, ETH Zurich, CSCS | 8B | SS | 64.4 tok/s | 5.4 GB | |
Gemma 4 E4B ITGoogle | 4B | SS | 50.3 tok/s | 6.9 GB | |
Gemma 3 4B ITGoogle | 4B | SS | 50.3 tok/s | 6.9 GB | |
Youtu-Parsing-Omnitencent | 5.33B | SS | 54.5 tok/s | 6.4 GB | |
Mistral 7B InstructMistral AI | 7B | SS | 54.4 tok/s | 6.4 GB | |
Ad Vast.aiAffordable on-demand GPU rentals for training and inference. Pick from thousands of hosts.Rent a GPU | |||||
Holo4-35B-A3BHcompany | 35B(3B active) | SS | 148.7 tok/s | 2.3 GB | |
PersonaPlex 7BNVIDIA | 7B | SS | 72.6 tok/s | 4.8 GB | |
Llama 2 7B ChatMeta | 7B | SS | 72.6 tok/s | 4.8 GB | |
LFM2.5-8B-A1BLiquid AI | 8.3B(1.5B active) | SS | 119.7 tok/s | 2.9 GB | |
Mellum2.1JetBrains | 12B(2.5B active) | SS | 171.1 tok/s | 2.0 GB | |
LFM2.5-2.6BLiquid AI | 2.69B | AA | 71.2 tok/s | 4.9 GB | |
MiniCPM-V 4.7 1Bopenbmb | 1.55B | AA | 74.8 tok/s | 4.6 GB | |
MiniCPM5-2Bopenbmb | 2.52B | AA | 75.5 tok/s | 4.6 GB | |
| AdTools for AI BuildersAll the hardware and tools you need to run private AI at your office or at home.Shop Now | |||||
DiffusionGemma 26B-A4BGoogle | 25.2B(3.8B active) | AA | 33.2 tok/s | 10.5 GB | |
Model Compatibility
Use the calculator to weigh this device against any other, model by model, with speed and fit scores.
Straight answers to the questions we hear most often.
Can't find what you're looking for? Book a Discovery Call
AliceAI-Foundation-80B-A3B-Base is the highest-rated model that runs on AMD Radeon RX 7700 XT at 4-bit, at about 41 tok/s. In total, AMD Radeon RX 7700 XT can run 204 of the 301 open models we track.
We size each model at 4-bit (Q4_K_M), add the KV cache for its context length and a fixed runtime overhead, then compare that to this device VRAM. Each model is graded on how comfortably it fits and how fast it should run on this card.
It is an estimate of decode speed in tokens per second at 4-bit, based on this device memory bandwidth and each model size. Real speed depends on your runtime, batch size, and context length, so use it to compare models, not as a hard promise.
Lower-bit quantization shrinks a model so more of them fit. These grades already assume 4-bit, which is the common local default. Going below 4-bit can squeeze a larger model in at some quality cost; going above needs more VRAM and may not fit.
Mixture-of-experts models only activate a fraction of their parameters at a time, so their runtime memory is closer to the active size than the headline size. We score fit on the active parameters, which is why some very large models still fit.
