Benchmark Results

Fraktur Adverts

Assesses models' capability to recognize and transcribe historical German Fraktur script, a Gothic typeface commonly used in German-language documents.

Dataset Description Result Overview Test Runs

This benchmark has been run 179 times. It uses cer metric.

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

Tested models: qwen/qwen3.5-9b, qwen3.5-27b, claude-sonnet-4-6, gpt-4.5-preview, gpt-5.3-codex, gpt-5-nano, qwen/qwen3.5-122b-a10b, qwen/qwen3-vl-8b-thinking, claude-3-5-sonnet-20241022, gpt-5.4-2026-03-05, gemini-2.5-flash-lite-preview-09-2025, qwen3.5-plus-2026-02-15, mistral-medium-2505, gpt-4.1-nano, qwen35-397b-a17b-fp8, gemini-2.5-pro-preview-05-06, claude-opus-4-1-20250805, gemini-3-pro-preview, GLM-4.5V-FP8, qwen/qwen3-vl-30b-a3b-instruct, gemini-3.1-flash-lite, gpt-4o, magistral-small-2509, gemini-2.5-flash-preview-04-17, gemini-1.5-pro, gemini-2.5-flash-preview-09-2025, qwen/qwen3.5-plus-02-15, claude-sonnet-4-20250514, claude-opus-4-20250514, qwen/qwen3.7-plus, o3, ministral-14b-2512, gemini-3.5-flash, claude-3-opus-20240229, claude-sonnet-4-5-20250929, google/gemma-4-31b-it, grok-4.3, mistral-medium-3.5, google/gemma-4-26b-a4b-it, claude-haiku-4-5-20251001, mistral-large-2512, gemini-2.5-pro, qwen/qwen3-vl-8b-instruct, gemini-2.5-flash, claude-opus-4-6, grok-4.20-0309-reasoning, gpt-4o-mini, qwen3.5-flash-2026-02-23, gemini-2.5-flash-lite, gemini-2.0-flash-lite, qwen/qwen3.6-plus, qwen3.5-397b-a17b, pixtral-large-2411, meta-llama/llama-4-scout, qwen/qwen3.5-397b-a17b, gemini-3-flash-preview, mistral-medium-2508, stepfun/step-3.7-flash, qwen/qwen3.5-27b, claude-opus-4-5-20251101, claude-opus-4-7, mistral-large-2411, gpt-5.5-2026-04-23, pixtral-12b, claude-3-7-sonnet-20250219, gemini-2.0-flash, qwen/qwen3.5-35b-a3b, gpt-4.1-mini, qwen3.5-122b-a10b, gemini-3.1-pro-preview, gpt-5.2-2025-12-11, magistral-medium-2509, claude-sonnet-5, qwen/qwen3.5-flash-02-23, gpt-5-mini, ministral-8b-2512, claude-fable-5, gemini-1.5-flash, gemini-exp-1206, x-ai/grok-4, gemini-3.1-flash-lite-preview, qwen3.5-35b-a3b, claude-opus-4-8, gemini-2.0-pro-exp-02-05, meta-llama/llama-4-maverick, gpt-5, gpt-4.1, mistral-small-2506, gpt-5.1-2025-11-13, gemini-2.5-pro-exp-03-25

• Type(s): book-page
• Benchmark task(s):  transcription
• Writing: printed
• Source creation (century): 18, 19
• Source Layout: prose
• Language(s): de