RX 7900 XT LLM benchmarks
131 results · 157 measurements · 9 sources
Device specifications
| Variant | Memory | Memory type | Bus | Bandwidth | FP16 matrix | INT8 matrix |
|---|---|---|---|---|---|---|
| RX 7900 XT ↗ | 20 GB | GDDR6 | 320-bit | 800 GB/s | 103 TFLOPS | 103 TOPS |
Manufacturer specifications. Shared system RAM is not dedicated VRAM. Unreported values remain —; measured LLM results retain their reported configuration.
Selected reported tests
24 of 131 resultsSource-reported results, selected across reports by documentation quality. Match configurations before comparing values.
| View ReportLocalMaxxing | Qwen3.6-27B | Token speed | — | 375 tok/sPrefill · per request | 60.6 tok/sGeneration · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
|---|---|---|---|---|---|---|
| View ReportGeneric LLM | Qwen3.5-9B | Token speed | — | 1,851 tok/sPrefill · per request | 83.82 tok/sGeneration · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | DeepSeek-R1-Distill-Llama-8B | Token speed | — | 2,671.84 tok/sPrefill · mean · per request | — | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View ReportServing benchmark | Huihui-Qwen3.8-27B-abliterated | Token speedladder | — | — | 35.04 tok/sGeneration · median · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Llama-2-7B | Token speed | — | 2,967.12 tok/sPrefill · mean · per request | — | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Qwen3.6-35B-A3B | Token speed | — | — | 109.73 tok/sGeneration · mean · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Llama-2-7B | Token speed | — | 3,098.38 tok/sPrefill · mean · per request | — | GPU count unknownUnknown · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Llama-2-7B | Token speed | — | 2,941.58 tok/sPrefill · mean · per request | — | GPU count unknownUnknown · 20 GB VRAM/GPUConfiguration
|
| View ReportServing benchmark | Qwen3.8-27B | Other LLM measurementsPython code1 concurrent request | 30.8 updates/sStream update rate · Per request | — | — | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View ReportLocalMaxxing | Nemotron-3.5-Lightning-30B-A3B | Token speed | — | 1,191.7 tok/sPrefill · per request | 116 tok/sGeneration · per request | 2 × RX 7900 XTMulti-GPU · split unknown · 20 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | Qwen3.5-9B | Token speed | — | 1,129 tok/sPrefill · mean · per request | 82.41 tok/sGeneration · mean · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | DeepSeek-R1-Distill-Llama-8B | Token speed | — | — | 80.9 tok/sGeneration · mean · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View ReportServing benchmark | Huihui-Qwen3.8-27B-abliterated | Token speedladder | — | — | 33.87 tok/sGeneration · median · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Llama-2-7B | Token speed | — | — | 116 tok/sGeneration · mean · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Qwen3.6-35B-A3B | Token speed | — | — | 108.98 tok/sGeneration · mean · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Llama-2-7B | Token speed | — | — | 116.15 tok/sGeneration · mean · per request | GPU count unknownUnknown · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Llama-2-7B | Token speed | — | — | 123.18 tok/sGeneration · mean · per request | GPU count unknownUnknown · 20 GB VRAM/GPUConfiguration
|
| View ReportServing benchmark | Qwen3.8-27B | Other LLM measurementsmmap prose1 concurrent request | 30.9 updates/sStream update rate · Per request | — | — | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View ReportLocalMaxxing | Nemotron-3.5-Lightning-30B-A3B | Token speed | — | 535.3 tok/sPrefill · per request | 115.6 tok/sGeneration · per request | 2 × RX 7900 XTMulti-GPU · split unknown · 20 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | Qwen3.5-9B | Token speed | — | 1,343 tok/sPrefill · per request | 68.53 tok/sGeneration · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | gpt-oss-20b | Token speed | — | 3,323.19 tok/sPrefill · mean · per request | — | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View ReportServing benchmark | Huihui-Qwen3.8-27B-abliterated | Token speedladder | — | 747.25 tok/sPrefill · median · per request | — | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Llama-2-7B | Token speed | — | 3,163.24 tok/sPrefill · mean · per request | — | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | Qwen3.6-35B-A3B | Token speed | — | — | 105.61 tok/sGeneration · mean · per request | 1 × RX 7900 XTSingle GPU · 20 GB VRAM/GPUConfiguration
|
Models and software in this collection
Model families: DeepSeek R1, GPT-OSS, Granite 3.0, Llama 2, Llama 3.1, Mistral, Qwen3.5, Qwen3.6, Qwen3.8
Engines: llama.cpp, qwen-kernel, vllm
Methodology and evidence standards · Download the research dataset