AMD-Pro-BenchCommunity

13,812 benchmarks in 32 GPU combinations documented

Phi-3.5-mini-instruct on AMD GPUs

22 results · 3 sources · 34 measurements

Explore all results

Grouped by reported model identity. Exact weights, quantizations and software can differ. How results are reviewed

Documented results

22 selected tests
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon——
94.45 tok/sGeneration · per request
1 × RX 9060 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11 Pro for Workstations
Engine
ONNX Runtime
Backend
DirectML
Harness
Procyon
Prompt / output
? / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2025-06-04
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon——
105.95 tok/sGeneration · per request
1 × RX 9070 GRESingle GPU · 12 GB VRAM/GPU
Configuration
Memory
12 GB VRAM/GPU
OS
Windows 11 Pro for Workstations
Engine
ONNX Runtime
Backend
DirectML
Harness
Procyon
Prompt / output
? / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-01
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191——
196.85 tok/sGeneration · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
39 / 512 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon——
139.19 tok/sGeneration · per request
1 × RX 9070Single GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11 Pro for Workstations
Engine
ONNX Runtime
Backend
DirectML
Harness
Procyon
Prompt / output
? / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2025-06-04
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191——
191.53 tok/sGeneration · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
136 / 512 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon——
144.47 tok/sGeneration · per request
1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11 Pro for Workstations
Engine
ONNX Runtime
Backend
DirectML
Harness
Procyon
Prompt / output
? / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2025-06-04
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191——
174.41 tok/sGeneration · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
660 / 128 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191——
123.37 tok/sGeneration · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
2016 / 512 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191——
121.72 tok/sGeneration · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
2080 / 512 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191——
125.95 tok/sGeneration · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
2023 / 256 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191——
125.95 tok/sGeneration · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
2042 / 256 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191——
151.4 tok/sGeneration · mean · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
? / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.19139 input · 512 output
320 msTime to first token · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
39 / 512 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191136 input · 512 output
260 msTime to first token · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
136 / 512 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191660 input · 128 output
470 msTime to first token · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
660 / 128 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.1912,016 input · 512 output
1,050 msTime to first token · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
2016 / 512 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.1912,080 input · 512 output
1,070 msTime to first token · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
2080 / 512 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.1912,023 input · 256 output
1,060 msTime to first token · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
2023 / 256 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.1912,042 input · 256 output
1,040 msTime to first token · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
2042 / 256 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191
2,267 pointsBenchmark score · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
? / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191
750 msTime to first token · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
? / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
View
ReportGeneric LLM		Phi-3.5-mini-instructQuant not reportedModel ref. ↗Procyon1.0.191
1,190 msModel load time · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Windows (inferred; version unreported) (inferred)
Engine
Procyon AI Text Generation / ONNXRuntime-DirectML
Backend
DirectML
Harness
Procyon
Prompt / output
? / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2026-06-26
Full record →
Quantizations and software

Quantizations: Not reported
Engines: ONNX Runtime, Procyon AI Text Generation / ONNXRuntime-DirectML

Explore all 22 results →

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