tau2-bench telecom results on AMD GPUs

3 results · 3 sources · 3 measurements

Scores from different versions, task subsets, harnesses and scoring rules are not directly comparable. How results are reviewed

Documented results

3 selected tests


View ReportGeneric LLM	gemma-4-26B-A4B-it-MXFP4-Quark-RDNA4MXFP4 (Quark; mixed BF16)Weights ↗	tau2-bench telecom114 telecom tasks6 concurrent requests	46.49 %Reward · 53 / 114	—	—	2 × R9700Tensor parallel · 32 GB VRAM/GPU Configuration Memory 32 GB VRAM/GPU OS Not reported Engine vLLM (Capicua25x RDNA4 port) Backend ROCm / HIP Harness tau2-bench Prompt / output ? / ? tokens Context Not reported Concurrency 6 Date 2026-09-03 Full record →
View ReportGeneric LLM	Muse-Glimmer-30B-MXFP4-Quark-RDNA4MXFP4 (Quark; mixed BF16)Weights ↗	tau2-bench telecom114 telecom tasks6 concurrent requests	84.21 %Reward · 96 / 114	—	—	2 × R9700Tensor parallel · 32 GB VRAM/GPU Configuration Memory 32 GB VRAM/GPU OS Not reported Engine vLLM (Capicua25x RDNA4 port) Backend ROCm / HIP Harness tau2-bench Prompt / output ? / ? tokens Context 1048576 Concurrency 6 Date Not reported Full record →
View ReportGeneric LLM	Ornith-1.5-35B-A3B-MXFP4-Quark-RDNA4MXFP4 weight-only (Quark; BF16 activations)Weights ↗	tau2-bench telecom114 telecom tasks	96.5 %Reward · 110 / 114	—	—	2 × R9700Tensor parallel · 32 GB VRAM/GPU Configuration Memory 32 GB VRAM/GPU OS Not reported Engine vLLM (Capicua25x RDNA4 port) Backend ROCm / HIP Harness tau2-bench Prompt / output ? / ? tokens Context Not reported Concurrency Not reported Date Not reported Full record →

Versions and harnesses

Versions: Not reported
Harnesses: tau2-bench

This is a selection across sources, not a ranking or a matched experiment. Reading benchmark reports

GPU benchmark pages: R9700