R9700 LLM benchmarks
7,243 results · 12,713 measurements · 133 sources
Device specifications
| Variant | Memory | Memory type | Bus | Bandwidth | FP16 matrix | INT8 matrix |
|---|---|---|---|---|---|---|
| R9700 ↗ | 32 GB | GDDR6 | 256-bit | 640 GB/s | 191 TFLOPS | 383 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 7,243 resultsSource-reported results, selected across reports by documentation quality. Match configurations before comparing values.
| View Reportllm-bench.io | Mellum2.1-12B-A2.5B-Thinking | llm-bench.io QualityFour sequential scenarios: agent workflow, code generation, role play, research | 76.04 / 100LLM judge · Mean | 1,622.33 tok/sPrefill · mean · per request | 232.1 tok/sGeneration · mean · per request | GPU count unknownUnknown · 32 GB VRAM/GPUConfiguration
|
|---|---|---|---|---|---|---|
| View ReportCustom evaluation | Qwen3.6-27B | ScaffoldBenchV4 · Six scenarios · SB-10, 15, 20, 30, 35, 491 concurrent request | 40 / 53Score · Custom rubric | — | — | 2 × R9700Layer split · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | Qwen3.6-27B | MBPPMBPP+ v0.2.0 · Base tests · 378 tasks1 concurrent request | 92.06 %Tests · pass@1 · 348 / 378 | — | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View ReportCustom evaluation | Devstral-24B | MMLU57 subjects · first item per subject1 concurrent request | 80.7 %Accuracy · 46 / 57 · Exact-choice parser | — | — | 2 × R9700Tensor parallel · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | Laguna-XS.2 | Tool-use ladderDepth rung 1 of7; 3 profile(s) · schema-v2 / agentic-code-log-v116,453 input · 1 concurrent request | 100 %Checks passed · 3 / 3 · Tool-use checks | — | — | 2 × R9700Tensor parallel · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | Qwen3.8-27B | Tool-use ladderDepth rung 1 of7; 1 profile(s) · schema-v2 / repeated-v116,784 input · 1 concurrent request | 100 %Checks passed · 1 / 1 · Tool-use checks | — | — | 2 × R9700Tensor parallel · 32 GB VRAM/GPUConfiguration
|
| View ReportCoding session | Qwen3.8-Flash-Next | Terminal-Bench-LocalCore19 v1.0.0 · Core19 full19-task computing subset1 concurrent request | 89.47 %pass@2 · retry failures · 17 / 19 · Executable verifiers | — | — | 2 × R9700Multi-GPU · split unknownConfiguration
|
| View ReportGeneric LLM | sentinel-qwen38-27b | JevBench public panel231 decisions0 output · 1 concurrent request | 87.9 %Accuracy · n = 231 | — | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | GLM-5.3-Flash-RFA-RFI8-8xR9700 | GPQA Diamond100-task subset; 94 answered, 6 empty32,000 configured context | 85.1 %Accuracy among answered tasks · 80 / 94 | — | — | 8 × R9700Tensor parallel · 32 GB VRAM/GPUConfiguration
|
| View ReportCoding session | Qwen3.6-27BPartial evaluation | SWE-bench Verified MiniVerified Mini: 50 tasks (25 Django + 25 Sphinx) | 74 %LLM judge · 37 / 50 66 %Exit-zero rate · 33 / 50 · Container test exit | — | — | 2 × R9700Multi-GPU · split unknownConfiguration
|
| View ReportCoding session | Qwen3.6-27BPartial evaluation | SWE-bench Verified MiniVerified Mini: 50 tasks (25 Django + 25 Sphinx) | 70 %LLM judge · 35 / 50 58 %Exit-zero rate · 29 / 50 · Container test exit | — | — | 1 × R9700Single GPUConfiguration
|
| View ReportCoding session | Qwen3.6-27BPartial evaluation | SWE-bench Verified MiniVerified Mini: 50 tasks (25 Django + 25 Sphinx) | 72 %LLM judge · 36 / 50 62 %Exit-zero rate · 31 / 50 · Container test exit | — | — | 1 × R9700Single GPUConfiguration
|
| View ReportCoding session | Qwen3.6-35B-A3BPartial evaluation | SWE-bench Verified MiniVerified Mini: 50 tasks (25 Django + 25 Sphinx) | 66 %LLM judge · 33 / 50 56 %Exit-zero rate · 28 / 50 · Container test exit | — | — | 1 × R9700Single GPUConfiguration
|
| View ReportCoding session | gemma-4-26B-A4B-itPartial evaluation | SWE-bench Verified MiniVerified Mini: 50 tasks (25 Django + 25 Sphinx) | 50 %LLM judge · 25 / 50 52 %Exit-zero rate · 26 / 50 · Container test exit | — | — | 1 × R9700Single GPUConfiguration
|
| View ReportGeneric LLM | gemma-4-E2B-it | LoRA trainingDataset not reported | 2.86 lossTraining loss · Loss · Lower is better | — | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View ReportCustom evaluation | gte-Qwen2-1.5B-instruct | SciFact300 queries | 78.53 %Score · n = 1 · nDCG@10 | — | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View ReportCustom evaluation | Qwen2.5-7B-Instruct | Unspecified evaluation512 configured context | 7.552 PPLScore · Lower is better | — | — | 2 × R9700Tensor parallel · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | Qwen2.5-Coder-32B-Instruct | LoRA trainingDataset not reported | 0.4101 lossTraining loss · Loss · Lower is better | — | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View ReportBetterBench | Qwen3.8-27B-PARO-MXFP4-v2 | WikiText256K configured context | 5.97 perplexityWord perplexity · Served perplexity · Lower is better | — | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | Qwen3.8-27B | WikiText16 chunks of512 tokens1 concurrent request | 7.2 perplexityWord perplexity · Perplexity · Lower is better | — | — | 1 × R9700Single GPU · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | gemma-4-26B-A4B-it-MXFP4-Quark-RDNA4 | tau2-bench telecom114 telecom tasks6 concurrent requests | 46.49 %Reward · 53 / 114 | — | — | 2 × R9700Tensor parallel · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | Qwen3.8-Flash-Next uncensored fine-tune (exact checkpoint unreported) | HumanEval164 problems | 87.8 %Reported pass@1 · n = 164 | — | — | 3 × R9700Tensor parallel · 32 GB VRAM/GPUConfiguration
|
| View ReportCustom evaluation | Ornith-1.5-35B-A3B-MXFP4-Quark-RDNA4 | GSM8K50 prompts per seed; flexible extraction | 90 %Accuracy · flexible extraction · Median | — | — | 2 × R9700Tensor parallel · 32 GB VRAM/GPUConfiguration
|
| View ReportGeneric LLM | Muse-Glimmer-30B-MXFP4-Quark-RDNA4 | tau2-bench telecom114 telecom tasks6 concurrent requests | 84.21 %Reward · 96 / 114 | — | — | 2 × R9700Tensor parallel · 32 GB VRAM/GPUConfiguration
|
Models and software in this collection
Model families: BGE, BGE Reranker v2, DeepSeek Coder V2, DeepSeek MoE, DeepSeek R1, DeepSeek V4, DeepSeek V4.1, Devstral, Devstral 2, Devstral Small 2, GLM 4.5, GLM 4.7, GLM 5.3, GPT-OSS, GTE Qwen2, Gemma 2, Gemma 3, Gemma 4, Granite 3.0, Laguna, Llama 2, Llama 3, Llama 3.1, Llama 3.2, MiniCPM4, Ministral 3, Mistral, Mistral Small 3.1, Mistral Small 3.2, Muse Glimmer, Nemotron 3, North Mini Code, Phi 3.5, Phi 4, Qwen2.5, Qwen3, Qwen3.5, Qwen3.6, Qwen3.8, TinyLlama
Engines: Affinity, Burn, ComfyUI / PyTorch (footnote wording), LLaMA-Factory, LM Studio, LM Studio / llama.cpp, Lemonade / llama.cpp, Ollama, Paiton + vLLM, Procyon AI Text Generation / ONNXRuntime-DirectML, PyTorch, R9V / vLLM, Radiance, Radiance standalone, SGLang, Strata, Transformers, Transformers / PEFT, custom, hipfire, llama.cpp, llama.cpp + Sentinel proxy, llama.cpp / ROCmFPX, llamafile, lucebox, public vLLM Radiance comparator, ranma.cpp, strata, vLLM, vLLM (Capicua25x RDNA4 port), vLLM (IronLLM Labs overlay), vLLM + r9700-stack, vLLM / Radiance, vLLM Radiance, vLLM Radiance MXFP4, vLLM-Radiance, vLLM_for_AMD, vllm
Methodology and evidence standards · Download the research dataset