Grouped by reported model identity. Exact weights, quantizations and software can differ. How results are reviewed
Documented results
24 selected tests| View Reportllm-bench.io | gpt-oss-20b | llm-bench.io QualityFour sequential scenarios: agent workflow, code generation, role play, research | 72.82 / 100LLM judge · Mean | 2,004.05 tok/sPrefill · mean · per request | 93.46 tok/sGeneration · mean · per request | GPU count unknownUnknown · 16 GB VRAM/GPUConfiguration
|
|---|---|---|---|---|---|---|
| View Reportllm-bench.io | gpt-oss-20b | llm-bench.io QualityFour sequential scenarios: agent workflow, code generation, role play, research | 77.86 / 100LLM judge · Mean | — | 63.47 tok/sGeneration · mean · per request | GPU count unknownUnknown · 24 GB VRAM/GPUConfiguration
|
| View ReportCustom evaluation | gpt-oss-20b | Local LLM custom suite4b · Math / math_basic | 100 %Score · 1 / 1 · Custom answer parser 90.22 tok/sThroughput · per request | — | — | 1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPUConfiguration
|
| View ReportLocalMaxxing | gpt-oss-20b | Token speed | — | 1,321.4 tok/sPrefill · median · unknown | 76.1 tok/sGeneration · median · unknown | 1 × Ryzen AI Max+ 395Single GPU · Unified memoryConfiguration
|
| View ReportLocalMaxxing | gpt-oss-20b | Token speed | — | 1,785.17 tok/sPrefill · unknown | 100.12 tok/sGeneration · unknown | 1 × RX 9060 XTSingle GPU · 16 GB VRAM/GPUConfiguration
|
| View ReportLocalMaxxing | gpt-oss-20b | Token speed | — | 3,089.13 tok/sPrefill · unknown | 165.8 tok/sGeneration · unknown | 1 × RX 7900 XTXSingle GPU · 24 GB VRAM/GPUConfiguration
|
| View ReportLocalMaxxing | gpt-oss-20b | Token speed | — | 2,122.92 tok/sPrefill · unknown | 138.83 tok/sGeneration · peak · unknown | 3 × R9700Multi-GPU · split unknown · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | gpt-oss-20b | Token speed | 71.43 tok/sInput + output · mean · per request | — | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | gpt-oss-20b | Token speed512 input tokens · 0 output tokens · 0 existing KV tokens | — | 1,786.17 tok/sPrefill · mean · per request | — | 1 × Ryzen AI Max+ 395Single GPU · Unified memoryConfiguration
|
| View Reportllama-bench | gpt-oss-20b | Token speed | — | 3,372 tok/sPrefill · per request | — | 1 × RX 9060 XTSingle GPU · 16 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | gpt-oss-20b | Token speed | — | — | 177.03 tok/sGeneration · mean · per request | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | gpt-oss-20b | Token speed | — | — | 154.68 tok/sGeneration · mean · per request | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | gpt-oss-20b | Token speed | — | — | 178.12 tok/sGeneration · mean · per request | 1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | gpt-oss-20b | Token speed | — | 3,884.21 tok/sPrefill · mean · per request | — | 1 × RX 7900 XTXSingle GPU · 24 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | gpt-oss-20b | Token speed | — | 5,529.9 tok/sPrefill · per request | — | 1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | gpt-oss-20b | Token speed2,048 input tokens · 16,384 existing KV tokens | — | 1,414.21 tok/sPrefill · mean · per request | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | gpt-oss-20b | Token speed2,048 input tokens · 16,384 existing KV tokens | — | 3,301.87 tok/sPrefill · mean · per request | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View Reportllama-bench | gpt-oss-20b | Token speed | — | 2,873.01 tok/sPrefill · per request | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | gpt-oss-20b | Token speed | 213.79 tok/sThroughput · unknown | — | — | GPU count unknownUnknown · 24 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | gpt-oss-20b | Token speed | — | ~1,750 tok/sPrefill · mean · per request | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View Reportllm-bench.io | gpt-oss-20b | llm-bench.io QualityFour sequential scenarios: agent workflow, code generation, role play, research | 71.99 / 100LLM judge · Mean | 1,992.35 tok/sPrefill · mean · per request | 93.25 tok/sGeneration · mean · per request | GPU count unknownUnknown · 16 GB VRAM/GPUConfiguration
|
| View Reportllm-bench.io | gpt-oss-20b | llm-bench.io QualityFour sequential scenarios: agent workflow, code generation, role play, research | 75.9 / 100LLM judge · Mean | — | 67.78 tok/sGeneration · mean · per request | GPU count unknownUnknown · 24 GB VRAM/GPUConfiguration
|
| View ReportCustom evaluation | gpt-oss-20b | Local LLM custom suite4b · Math / math_discount | 100 %Score · 1 / 1 · Custom answer parser 77.35 tok/sThroughput · per request | — | — | 1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPUConfiguration
|
| View ReportLocalMaxxing | gpt-oss-20b | Token speed | — | 1,326.6 tok/sPrefill · median · unknown | 74.2 tok/sGeneration · median · unknown | 1 × Ryzen AI Max+ 395Single GPU · Unified memoryConfiguration
|
Quantizations and software
Quantizations: F16, MXFP4, MXFP4_MOE, Q4_K_M, Q8_0
Engines: llama.cpp, LM Studio, Ollama, vllm
This is a selection across sources, not a ranking or a matched experiment. Reading benchmark reports
GPU benchmark pages: R9700 · RX 9070 XT · RX 9070 · RX 9070 GRE · RX 9060 XT · RX 7900 XTX · RX 7900 XT · RX 7900 GRE · RX 7800 XT · Ryzen AI Max+ 395