PQID-Bench Interactive Explorer

Frozen v1.0.0 evidence, rendered from archived evaluation records. Hover, zoom, isolate traces, and download SVG views without contacting a model provider.

Models21
Prompts154
Execution91.22%
Assembly91.03%
Signature52.66%
ES-Gap38.56 pp

Measurement ladder

Scroll horizontally to inspect the complete chart.

Model profiles

Scroll horizontally to inspect the complete chart.

Recovery matrix

Scroll horizontally to inspect the complete chart.

Repeatability

Scroll horizontally to inspect the complete chart.

Provider aggregates

Scroll horizontally to inspect the complete chart.

Model-level data

ModelProvider routeExecution AssemblySignatureOrdered ParameterSignature AC1
GPT-5.6 SolOpenAI99.35%99.35%62.99%58.44%57.14%1.000
Claude Fable 5Anthropic98.05%98.05%62.99%58.44%57.79%0.947
Gemini 3.1 Pro PreviewGoogle96.75%96.75%61.04%55.84%55.19%1.000
GPT-5.4 miniOpenAI98.05%97.40%60.39%56.49%55.19%0.945
GPT-5.5OpenAI97.40%97.40%60.39%55.84%55.19%0.982
Claude Opus 4.8Anthropic100.00%99.35%59.74%56.49%55.19%1.000
Claude Sonnet 4.6Anthropic99.35%99.35%59.09%54.55%53.90%0.946
DeepSeek V4 ProDeepSeek91.56%91.56%59.09%56.49%54.55%0.816
Codestral 25.01GitHub Models93.51%93.51%55.84%50.65%48.05%0.944
Gemini 2.5 ProGoogle88.31%88.31%53.90%50.00%50.00%0.833
GPT-OSS 120BGroq96.75%96.10%53.25%50.00%48.70%0.636
DeepSeek V4 FlashDeepSeek88.96%88.96%52.60%50.00%49.35%0.815
GPT-OSS 20BGroq92.86%92.21%50.65%47.40%45.45%0.553
Qwen3-Coder-NextHugging Face85.71%85.71%50.65%44.81%44.81%0.907
Llama 4 MaverickGitHub Models94.16%93.51%48.70%46.10%46.10%0.629
Mistral Small 3.2 24BOpenRouter90.26%90.26%48.70%43.51%42.21%0.889
Llama 3.3 70BGroq93.51%93.51%46.10%40.91%40.26%0.853
Qiskit Mistral 3.2 24BHugging Face89.61%88.96%44.81%42.21%41.56%0.572
Llama 4 ScoutGroq82.47%82.47%44.81%39.61%38.96%0.835
Llama 3.1 8BGroq81.17%81.17%35.06%33.77%33.77%0.802
Qwen3 32BGroq57.79%57.79%35.06%31.82%29.87%0.661