Benchmark Results

Magazine Pages - Anglo Swiss Trade Review

Examines a model's ability to extract bounding boxes of advertisements from magazine pages.

Dataset Description Result Overview Test Runs

This benchmark has been run 64 times. It uses f1 metric.

Tested providers: x-ai, contour_local, openrouter, genai, openai, anthropic, mistral, alibaba

Tested models: claude-haiku-4-5-20251001, google/gemma-4-26b-a4b-it, qwen3.5-397b-a17b, qwen/qwen3.5-122b-a10b, gemini-2.0-flash, claude-opus-4-6, mistral-large-2411, qwen3.5-35b-a3b, claude-sonnet-4-5-20250929, mistral-small-2506, grok-4.20-0309-reasoning, gpt-5-nano, gemini-2.0-flash-lite, mistral-medium-2505, qwen/qwen3.5-flash-02-23, claude-opus-4-20250514, o3, opencv-contour, meta-llama/llama-4-maverick, gpt-4o-mini, ministral-14b-2512, qwen/qwen3-vl-8b-instruct, gemini-2.5-flash-lite, mistral-medium-2508, gemini-2.5-flash-lite-preview-09-2025, qwen3.5-122b-a10b, claude-opus-4-5-20251101, magistral-medium-2509, claude-opus-4-1-20250805, gpt-5.5-2026-04-23, gpt-5.4-2026-03-05, gemini-3.1-pro-preview, gpt-5, gpt-5.2-2025-12-11, gemini-2.5-pro, qwen3.5-27b, qwen/qwen3.5-27b, claude-sonnet-4-20250514, gemini-3.1-flash-lite-preview, qwen/qwen3-vl-8b-thinking, gpt-5-mini, claude-opus-4-7, magistral-small-2509, gpt-4.1, qwen3.5-flash-2026-02-23, gemini-2.5-flash, qwen/qwen3.5-35b-a3b, qwen/qwen3.5-397b-a17b, qwen/qwen3.5-9b, gpt-4.1-mini, qwen/qwen3-vl-30b-a3b-instruct, qwen3.5-plus-2026-02-15, claude-sonnet-4-6, qwen/qwen3.6-plus, gpt-4.1-nano, gpt-4o, gemini-3-flash-preview, pixtral-large-2411, mistral-large-2512, ministral-8b-2512, gpt-5.3-codex, google/gemma-4-31b-it, qwen/qwen3.5-plus-02-15, gpt-5.1-2025-11-13

• Type(s): newspaper-page
• Benchmark task(s):  document-understanding
• Writing: printed
• Source creation (century): 20
• Source Layout: prose, columns
• Language(s): en