{"$schema":"https://www.openui.com/benchmarks/data.schema.json","schema_version":"1.0.0","dataset":{"name":"OpenUI Generative UI Benchmark","canonical_url":"https://www.openui.com/benchmarks","version":"v1","benchmark_updated":"2026-08-18","model_board_updated":"2026-08-25","benchmark_updated_label":"25 Aug 2026","currency":"USD","pricing_basis":"Measured tokens at provider list prices unless cost_type says otherwise.","maintainer":"OpenUI by Thesys","first_party_disclosure":"OpenUI built and maintains this first-party benchmark."},"definitions":{"valid":"Parses, has a root, resolves every reference, has no orphaned or invented components, uses valid props, is not truncated, and meets the component-count floor.","render_success":"The format's own shipped renderer produced a non-blank screen.","cost_per_task_usd":"Measured cost of one generated screen, in USD.","cost_type":{"provider-list-price":"Comparable API price calculated from measured tokens and list prices.","free":"The source reports zero marginal API price.","self-hosted":"No comparable API price; cost_per_task_usd is null, never zero."},"pareto_frontier":"True when no other comparably priced model is at least as valid and strictly better on cost or validity.","family_order":"One-based catalog order within a named model family; null for singleton models. Families are not connected in the chart.","connected_family":"Whether members of a model family are connected by a chart line. False for every row in this chart version.","default_selected":"Whether the model is selected in the chart's opening view, which is every model at or above 70% structural validity.","generation_condition":"Four generations per brief, a 16,384-token output ceiling, temperature 0.7 where the provider accepts it, and minimal or no reasoning. Anthropic runs use the model default temperature because that API rejects the benchmark temperature setting for the tested model.","scorer_regime":"Every row is scored by one build of the shipped OpenUI parser, lang-core 0.2.16, whose parser validates enum and scalar prop values. The website and the benchmark repository are on the same regime; raw outputs and per-run verdicts are committed for independent rescoring."},"scope":{"briefs":46,"formats":3,"comparison_models":6,"model_board_models":30,"runs_per_format":1104,"total_scored_runs":3312},"provenance":{"format_comparison":{"evidence_status":"published","scorer_regime":"lang-core 0.2.16","raw_results":"https://github.com/thesysdev/generative-ui-bench/tree/main/results","raw_outputs":"https://github.com/thesysdev/generative-ui-bench/tree/main/raw","harness":"https://github.com/thesysdev/generative-ui-bench","briefs":"https://github.com/thesysdev/generative-ui-bench/tree/main/briefs/briefs.ts"},"model_board":{"evidence_status":"summary-only","note":"The model-board measurements are published here as summary rows. Raw generations and per-run verdicts for every row are committed in the benchmark repository for independent rescoring.","raw_results":null}},"model_families":[{"id":"gpt-5-6","label":"GPT-5.6","provider":"OpenAI","models":["gpt-5-6-sol","gpt-5-6-terra","gpt-5-6-luna"]},{"id":"claude","label":"Claude","provider":"Anthropic","models":["claude-opus-5","claude-opus-4-8","claude-sonnet-5","claude-sonnet-4-6"]},{"id":"gemini-flash","label":"Gemini Flash","provider":"Google","models":["gemini-3-7-flash","gemini-3-6-flash","gemini-3-5-flash-lite"]},{"id":"gemma-4","label":"Gemma 4","provider":"Google","models":["gemma-4-31b","gemma-4-26b-a4b"]},{"id":"qwen-3-8","label":"Qwen3.8","provider":"Alibaba","models":["qwen-3-8-2-4t","qwen-3-8-27b"]},{"id":"qwen-3-6","label":"Qwen3.6","provider":"Alibaba","models":["qwen-3-6-27b","qwen-3-6-35b-a3b"]},{"id":"deepseek-v4","label":"DeepSeek V4","provider":"DeepSeek","models":["deepseek-v4-pro","deepseek-v4-flash"]},{"id":"inkling","label":"Inkling","provider":"Thinking Machines","models":["inkling","inkling-small"]}],"model_board":[{"model_id":"grok-4-6","model_name":"Grok 4.6","provider":"xAI","family_id":"grok-4-6","family_name":"Grok 4.6","family_order":null,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":99.5,"cost_per_task_usd":0.01847826086956522,"cost_per_46_screen_pass_usd":0.85,"cost_type":"provider-list-price","serving":null,"pareto_frontier":true,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"gpt-5-6-sol","model_name":"GPT-5.6 Sol","provider":"OpenAI","family_id":"gpt-5-6","family_name":"GPT-5.6","family_order":1,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":99.5,"cost_per_task_usd":0.04760869565217391,"cost_per_46_screen_pass_usd":2.19,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"claude-opus-4-8","model_name":"Claude Opus 4.8","provider":"Anthropic","family_id":"claude","family_name":"Claude","family_order":2,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":98.9,"cost_per_task_usd":0.04934782608695652,"cost_per_46_screen_pass_usd":2.27,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"gemini-3-7-flash","model_name":"Gemini 3.7 Flash","provider":"Google","family_id":"gemini-flash","family_name":"Gemini Flash","family_order":1,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":98.9,"cost_per_task_usd":0.01,"cost_per_46_screen_pass_usd":0.46,"cost_type":"provider-list-price","serving":null,"pareto_frontier":true,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"claude-sonnet-5","model_name":"Claude Sonnet 5","provider":"Anthropic","family_id":"claude","family_name":"Claude","family_order":3,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":98.4,"cost_per_task_usd":0.020217391304347826,"cost_per_46_screen_pass_usd":0.93,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"gpt-5-6-terra","model_name":"GPT-5.6 Terra","provider":"OpenAI","family_id":"gpt-5-6","family_name":"GPT-5.6","family_order":2,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":98.4,"cost_per_task_usd":0.0241304347826087,"cost_per_46_screen_pass_usd":1.11,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"kimi-k3","model_name":"Kimi K3","provider":"Moonshot","family_id":"kimi-k3","family_name":"Kimi K3","family_order":null,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":96.2,"cost_per_task_usd":0.03326086956521739,"cost_per_46_screen_pass_usd":1.53,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"claude-opus-5","model_name":"Claude Opus 5","provider":"Anthropic","family_id":"claude","family_name":"Claude","family_order":1,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":96.2,"cost_per_task_usd":0.07739130434782608,"cost_per_46_screen_pass_usd":3.56,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"muse-spark-1-2","model_name":"Muse Spark 1.2","provider":"Meta","family_id":"muse-spark-1-2","family_name":"Muse Spark 1.2","family_order":null,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":96.2,"cost_per_task_usd":0.015652173913043476,"cost_per_46_screen_pass_usd":0.72,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"claude-sonnet-4-6","model_name":"Claude Sonnet 4.6","provider":"Anthropic","family_id":"claude","family_name":"Claude","family_order":4,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":92.9,"cost_per_task_usd":0.045,"cost_per_46_screen_pass_usd":2.07,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"qwen-3-8-2-4t","model_name":"Qwen3.8 2.4T","provider":"Alibaba","family_id":"qwen-3-8","family_name":"Qwen3.8","family_order":1,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":91.8,"cost_per_task_usd":0.016956521739130436,"cost_per_46_screen_pass_usd":0.78,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"glm-5-3","model_name":"GLM-5.3","provider":"Zhipu","family_id":"glm-5-3","family_name":"GLM-5.3","family_order":null,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":90.8,"cost_per_task_usd":0.011956521739130435,"cost_per_46_screen_pass_usd":0.55,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"inkling-small","model_name":"Inkling Small","provider":"Thinking Machines","family_id":"inkling","family_name":"Inkling","family_order":2,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":88.6,"cost_per_task_usd":0.003260869565217391,"cost_per_46_screen_pass_usd":0.15,"cost_type":"provider-list-price","serving":null,"pareto_frontier":true,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"deepseek-v4-flash","model_name":"DeepSeek V4 Flash","provider":"DeepSeek","family_id":"deepseek-v4","family_name":"DeepSeek V4","family_order":2,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":85.8,"cost_per_task_usd":0.0004347826086956522,"cost_per_46_screen_pass_usd":0.02,"cost_type":"provider-list-price","serving":null,"pareto_frontier":true,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"deepseek-v4-pro","model_name":"DeepSeek V4 Pro","provider":"DeepSeek","family_id":"deepseek-v4","family_name":"DeepSeek V4","family_order":1,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":84.2,"cost_per_task_usd":0.008913043478260869,"cost_per_46_screen_pass_usd":0.41,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"gpt-5-6-luna","model_name":"GPT-5.6 Luna","provider":"OpenAI","family_id":"gpt-5-6","family_name":"GPT-5.6","family_order":3,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":83.7,"cost_per_task_usd":0.002173913043478261,"cost_per_46_screen_pass_usd":0.1,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"qwen-3-8-27b","model_name":"Qwen3.8 27B","provider":"Alibaba","family_id":"qwen-3-8","family_name":"Qwen3.8","family_order":2,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":78.8,"cost_per_task_usd":0.005434782608695652,"cost_per_46_screen_pass_usd":0.25,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"gemini-3-6-flash","model_name":"Gemini 3.6 Flash","provider":"Google","family_id":"gemini-flash","family_name":"Gemini Flash","family_order":2,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":78.8,"cost_per_task_usd":0.009130434782608695,"cost_per_46_screen_pass_usd":0.42,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"gemini-3-5-flash-lite","model_name":"Gemini 3.5 Flash-Lite","provider":"Google","family_id":"gemini-flash","family_name":"Gemini Flash","family_order":3,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":78.3,"cost_per_task_usd":0.004130434782608695,"cost_per_46_screen_pass_usd":0.19,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"inkling","model_name":"Inkling","provider":"Thinking Machines","family_id":"inkling","family_name":"Inkling","family_order":1,"connected_family":false,"default_selected":true,"format":"openui","validity_score_percent":73.4,"cost_per_task_usd":0.008043478260869565,"cost_per_46_screen_pass_usd":0.37,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"qwen-3-6-27b","model_name":"Qwen3.6 27B","provider":"Alibaba","family_id":"qwen-3-6","family_name":"Qwen3.6","family_order":1,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":68.5,"cost_per_task_usd":0.006304347826086956,"cost_per_46_screen_pass_usd":0.29,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"qwen-3-6-35b-a3b","model_name":"Qwen3.6 35B-A3B","provider":"Alibaba","family_id":"qwen-3-6","family_name":"Qwen3.6","family_order":2,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":61.4,"cost_per_task_usd":0.0017391304347826088,"cost_per_46_screen_pass_usd":0.08,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"gemma-4-31b","model_name":"Gemma 4 31B","provider":"Google","family_id":"gemma-4","family_name":"Gemma 4","family_order":1,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":46.7,"cost_per_task_usd":0.0008695652173913044,"cost_per_46_screen_pass_usd":0.04,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"phi-4","model_name":"Phi-4","provider":"Microsoft","family_id":"phi-4","family_name":"Phi-4","family_order":null,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":44,"cost_per_task_usd":0.0004347826086956522,"cost_per_46_screen_pass_usd":0.02,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"gemma-4-26b-a4b","model_name":"Gemma 4 26B-A4B","provider":"Google","family_id":"gemma-4","family_name":"Gemma 4","family_order":2,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":29.9,"cost_per_task_usd":0.0006521739130434782,"cost_per_46_screen_pass_usd":0.03,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"ministral-8b","model_name":"Ministral 8B","provider":"Mistral","family_id":"ministral-8b","family_name":"Ministral 8B","family_order":null,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":27.2,"cost_per_task_usd":0.0008695652173913044,"cost_per_46_screen_pass_usd":0.04,"cost_type":"provider-list-price","serving":null,"pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"granite-4-1-8b","model_name":"Granite 4.1 8B","provider":"IBM","family_id":"granite-4-1-8b","family_name":"Granite 4.1 8B","family_order":null,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":14.7,"cost_per_task_usd":0.0002173913043478261,"cost_per_46_screen_pass_usd":0.01,"cost_type":"provider-list-price","serving":null,"pareto_frontier":true,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"lfm-2-5-2-6b","model_name":"LFM 2.5 2.6B","provider":"Liquid","family_id":"lfm-2-5-2-6b","family_name":"LFM 2.5 2.6B","family_order":null,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":3.3,"cost_per_task_usd":0,"cost_per_46_screen_pass_usd":0,"cost_type":"free","serving":null,"pareto_frontier":true,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"diffusion-gemma-26b-a4b","model_name":"DiffusionGemma 26B-A4B","provider":"Google","family_id":"diffusion-gemma-26b-a4b","family_name":"DiffusionGemma 26B-A4B","family_order":null,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":13,"cost_per_task_usd":null,"cost_per_46_screen_pass_usd":null,"cost_type":"self-hosted","serving":"Self-hosted · A100 FP8","pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"},{"model_id":"ling-3-tiny","model_name":"Ling 3.0 Tiny","provider":"InclusionAI","family_id":"ling-3-tiny","family_name":"Ling 3.0 Tiny","family_order":null,"connected_family":false,"default_selected":false,"format":"openui","validity_score_percent":9.8,"cost_per_task_usd":null,"cost_per_46_screen_pass_usd":null,"cost_type":"self-hosted","serving":"Self-hosted · llama.cpp Q8","pareto_frontier":false,"briefs":46,"generations_per_brief":4,"dataset_updated":"2026-08-25"}],"format_comparison":[{"model_id":"sol","model_name":"GPT-5.6 Sol","provider":"OpenAI","format_id":"openui","format_name":"OpenUI","valid_runs":183,"rendered_runs":184,"total_runs":184,"validity_score_percent":99.45652173913044,"render_rate_percent":100,"cost_per_46_screen_pass_usd":3.08},{"model_id":"sol","model_name":"GPT-5.6 Sol","provider":"OpenAI","format_id":"a2ui","format_name":"A2UI","valid_runs":177,"rendered_runs":179,"total_runs":184,"validity_score_percent":96.19565217391305,"render_rate_percent":97.28260869565217,"cost_per_46_screen_pass_usd":6.84},{"model_id":"sol","model_name":"GPT-5.6 Sol","provider":"OpenAI","format_id":"jsonRender","format_name":"json-render","valid_runs":152,"rendered_runs":184,"total_runs":184,"validity_score_percent":82.6086956521739,"render_rate_percent":100,"cost_per_46_screen_pass_usd":6.3},{"model_id":"opus","model_name":"Claude Opus 4.8","provider":"Anthropic","format_id":"openui","format_name":"OpenUI","valid_runs":182,"rendered_runs":184,"total_runs":184,"validity_score_percent":98.91304347826086,"render_rate_percent":100,"cost_per_46_screen_pass_usd":2.27},{"model_id":"opus","model_name":"Claude Opus 4.8","provider":"Anthropic","format_id":"a2ui","format_name":"A2UI","valid_runs":183,"rendered_runs":184,"total_runs":184,"validity_score_percent":99.45652173913044,"render_rate_percent":100,"cost_per_46_screen_pass_usd":5.85},{"model_id":"opus","model_name":"Claude Opus 4.8","provider":"Anthropic","format_id":"jsonRender","format_name":"json-render","valid_runs":161,"rendered_runs":184,"total_runs":184,"validity_score_percent":87.5,"render_rate_percent":100,"cost_per_46_screen_pass_usd":4.88},{"model_id":"kimi","model_name":"Kimi K3","provider":"Moonshot","format_id":"openui","format_name":"OpenUI","valid_runs":177,"rendered_runs":184,"total_runs":184,"validity_score_percent":96.19565217391305,"render_rate_percent":100,"cost_per_46_screen_pass_usd":1.53},{"model_id":"kimi","model_name":"Kimi K3","provider":"Moonshot","format_id":"a2ui","format_name":"A2UI","valid_runs":176,"rendered_runs":179,"total_runs":184,"validity_score_percent":95.65217391304348,"render_rate_percent":97.28260869565217,"cost_per_46_screen_pass_usd":3.16},{"model_id":"kimi","model_name":"Kimi K3","provider":"Moonshot","format_id":"jsonRender","format_name":"json-render","valid_runs":131,"rendered_runs":184,"total_runs":184,"validity_score_percent":71.19565217391305,"render_rate_percent":100,"cost_per_46_screen_pass_usd":2.91},{"model_id":"gemini","model_name":"Gemini 3.7 Flash","provider":"Google","format_id":"openui","format_name":"OpenUI","valid_runs":182,"rendered_runs":184,"total_runs":184,"validity_score_percent":98.91304347826086,"render_rate_percent":100,"cost_per_46_screen_pass_usd":0.42},{"model_id":"gemini","model_name":"Gemini 3.7 Flash","provider":"Google","format_id":"a2ui","format_name":"A2UI","valid_runs":173,"rendered_runs":178,"total_runs":184,"validity_score_percent":94.02173913043478,"render_rate_percent":96.73913043478261,"cost_per_46_screen_pass_usd":0.95},{"model_id":"gemini","model_name":"Gemini 3.7 Flash","provider":"Google","format_id":"jsonRender","format_name":"json-render","valid_runs":171,"rendered_runs":184,"total_runs":184,"validity_score_percent":92.93478260869566,"render_rate_percent":100,"cost_per_46_screen_pass_usd":0.85},{"model_id":"qwen","model_name":"Qwen3.8 2.4T","provider":"Alibaba","format_id":"openui","format_name":"OpenUI","valid_runs":169,"rendered_runs":183,"total_runs":184,"validity_score_percent":91.84782608695652,"render_rate_percent":99.45652173913044,"cost_per_46_screen_pass_usd":0.81},{"model_id":"qwen","model_name":"Qwen3.8 2.4T","provider":"Alibaba","format_id":"a2ui","format_name":"A2UI","valid_runs":168,"rendered_runs":168,"total_runs":184,"validity_score_percent":91.30434782608695,"render_rate_percent":91.30434782608695,"cost_per_46_screen_pass_usd":1.9},{"model_id":"qwen","model_name":"Qwen3.8 2.4T","provider":"Alibaba","format_id":"jsonRender","format_name":"json-render","valid_runs":149,"rendered_runs":180,"total_runs":184,"validity_score_percent":80.97826086956522,"render_rate_percent":97.82608695652173,"cost_per_46_screen_pass_usd":1.46},{"model_id":"muse","model_name":"Muse Spark 1.2","provider":"Meta","format_id":"openui","format_name":"OpenUI","valid_runs":177,"rendered_runs":184,"total_runs":184,"validity_score_percent":96.19565217391305,"render_rate_percent":100,"cost_per_46_screen_pass_usd":0.71},{"model_id":"muse","model_name":"Muse Spark 1.2","provider":"Meta","format_id":"a2ui","format_name":"A2UI","valid_runs":178,"rendered_runs":179,"total_runs":184,"validity_score_percent":96.73913043478261,"render_rate_percent":97.28260869565217,"cost_per_46_screen_pass_usd":1.38},{"model_id":"muse","model_name":"Muse Spark 1.2","provider":"Meta","format_id":"jsonRender","format_name":"json-render","valid_runs":150,"rendered_runs":184,"total_runs":184,"validity_score_percent":81.52173913043478,"render_rate_percent":100,"cost_per_46_screen_pass_usd":1.34}],"format_summary":[{"format_id":"openui","format_name":"OpenUI","validity_score_percent":96.92028985507247,"render_rate_percent":99.90942028985508,"blank_screens":1,"runs":1104,"system_prompt_tokens":5031,"mean_output_tokens_per_screen":1362,"estimated_stream_seconds_at_50_tps":27.24},{"format_id":"a2ui","format_name":"A2UI","validity_score_percent":95.56159420289855,"render_rate_percent":96.64855072463769,"blank_screens":37,"runs":1104,"system_prompt_tokens":12610,"mean_output_tokens_per_screen":2823,"estimated_stream_seconds_at_50_tps":56.46},{"format_id":"jsonRender","format_name":"json-render","validity_score_percent":82.78985507246377,"render_rate_percent":99.63768115942028,"blank_screens":4,"runs":1104,"system_prompt_tokens":7651,"mean_output_tokens_per_screen":3258,"estimated_stream_seconds_at_50_tps":65.16}],"complexity_bands":[{"id":"b1","band":"2–3","briefs":10,"runs":240,"openui":100,"a2ui":99.58333333333333,"jsonRender":89.58333333333334},{"id":"b2","band":"4–6","briefs":10,"runs":240,"openui":99.58333333333333,"a2ui":98.75,"jsonRender":81.66666666666667},{"id":"b3","band":"7–9","briefs":10,"runs":240,"openui":98.75,"a2ui":94.16666666666667,"jsonRender":85.41666666666666},{"id":"b4","band":"11–13","briefs":8,"runs":192,"openui":91.14583333333334,"a2ui":91.66666666666666,"jsonRender":68.22916666666666},{"id":"b5","band":"16–18","briefs":8,"runs":192,"openui":90.625,"a2ui":93.22916666666666,"jsonRender":71.875}],"production_repair":{"triggerRate":7,"repairedShare":88,"userVisibleShare":0.9,"stages":[{"id":"all","label":"All generations","share":100,"fix":"Parser fixes syntax issues first","llm":false},{"id":"failed","label":"Fail validation","share":7,"fix":"Structural issues like dangling refs and bad enums","llm":false},{"id":"broken","label":"Reach a user broken","share":0.9,"fix":"88% are repaired via incremental editing","llm":true}],"source_scope":"Recent OpenUI Cloud production traffic; not benchmark runs."},"methodology":"https://www.openui.com/benchmarks/methodology","agent_answers":[{"id":"benchmark-scope","question":"How large is the benchmark?","answer":"46 interface briefs are used. The format comparison covers 6 models and 3 formats for 3,312 scored runs; the OpenUI model board covers 30 models.","evidence_paths":["scope","model_board","format_comparison"]},{"id":"valid-definition","question":"What counts as a valid generation?","answer":"A generation must parse, have a root, resolve every reference, contain no orphaned or invented components, use valid props, avoid truncation, and meet the component-count floor.","evidence_paths":["definitions.valid"]},{"id":"highest-model-validity","question":"Which models have the highest OpenUI validity score?","answer":"Grok 4.6 and GPT-5.6 Sol tie for the highest measured validity score at 99.5%. Cost determines which of these is Pareto-efficient; a high score alone does not imply frontier membership.","evidence_paths":["model_board"]},{"id":"openui-format-result","question":"How did OpenUI perform in the format comparison?","answer":"OpenUI achieved 96.9% structural validity and 99.9% render success across 1,104 runs, with 1 blank screen.","evidence_paths":["format_summary"]},{"id":"prompt-token-result","question":"Which format used the fewest system-prompt tokens?","answer":"OpenUI used the fewest system-prompt tokens at 5,031.","evidence_paths":["format_summary"]},{"id":"self-hosted-cost","question":"Does a zero or missing cost mean a self-hosted model is free?","answer":"No. Self-hosted models have no comparable API list price, so their cost is null. A numeric zero is reserved for a source reporting zero marginal API price.","evidence_paths":["definitions.cost_type","model_board"]},{"id":"pareto-meaning","question":"What does Pareto frontier mean in this benchmark?","answer":"A priced model is on the frontier when no other comparably priced model is at least as valid while being strictly better on cost or validity. Frontier membership is a cost-quality trade-off, not a general model ranking.","evidence_paths":["definitions.pareto_frontier","model_board"]},{"id":"chart-views","question":"What do the two tabs at the top of the benchmark show?","answer":"Model comparison shows OpenUI structural validity and cost across ${MODEL_BOARD_SIZE} models as provider-coloured dots with one Pareto frontier. Models scoring below 70% structural validity are deselected by default but remain in the data; both axes rescale to whatever is selected, so selecting a lower-scoring model extends the vertical scale down to reach it. Format comparison shows structural validity and cost across OpenUI, A2UI, and json-render for six models. Exact values for both views are present in the page's server-rendered chart data tables and machine-readable distributions.","evidence_paths":["model_board","model_families","format_comparison","distributions"]}],"distributions":{"schema":"https://www.openui.com/benchmarks/data.schema.json","json":"https://www.openui.com/benchmarks/data.json","csv":"https://www.openui.com/benchmarks/data.csv","markdown":"https://www.openui.com/benchmarks/agent.md","raw_results":"https://github.com/thesysdev/generative-ui-bench/tree/main/results","harness":"https://github.com/thesysdev/generative-ui-bench","briefs":"https://github.com/thesysdev/generative-ui-bench/tree/main/briefs/briefs.ts","language_benchmark":"https://www.openui.com/benchmarks/language","framework_benchmark":"https://www.openui.com/benchmarks/framework"}}