Model card
Qianfan-OCR.
Baiduopen-sourceOCR / document VLMcheck model card4 current SOTA
Added from Papers With Code public API dump for OCRBench.
§ 02 · Benchmarks
Every benchmark Qianfan-OCR has a recorded score for.
| # | Benchmark | Area · Task | Metric | Value | Rank | Date | Source |
|---|---|---|---|---|---|---|---|
| 01 | OCRBench v2 | — | overall-zh-private | 60.8% | #3 | 2026-04-20 | source ↗ |
| 02 | OCRBench v2 | — | overall-en-private | 56.0% | #9 | 2026-04-20 | source ↗ |
| 03 | OmniDocBench | Vision & Documents · Document Parsing | formula-cdm | 92.4% | #1 | 2026-04-20 | source ↗ |
| 04 | OmniDocBench | Vision & Documents · Document Parsing | table-teds | 91.0% | #2 | 2026-04-20 | source ↗ |
| 05 | OmniDocBench | Vision & Documents · Document Parsing | text-edit | 0.0% | #1 | 2026-04-20 | source ↗ |
| 06 | olmOCR-Bench | Vision & Documents · Document Parsing | base | 99.6% | #3 | 2026-04-20 | source ↗ |
| 07 | olmOCR-Bench | Vision & Documents · Document Parsing | multi-column | 92.2% | #1 | 2026-04-20 | source ↗ |
| 08 | olmOCR-Bench | Vision & Documents · Document Parsing | tables | 81.6% | #5 | 2026-04-20 | source ↗ |
| 09 | olmOCR-Bench | Vision & Documents · Document Parsing | long-tiny-text | 80.4% | #4 | 2026-04-20 | source ↗ |
| 10 | olmOCR-Bench | Vision & Documents · Document Parsing | arxiv | 80.1% | #5 | 2026-04-20 | source ↗ |
| 11 | olmOCR-Bench | Vision & Documents · Document Parsing | old-scans | 73.1% | #1 | 2026-04-20 | source ↗ |
| 12 | olmOCR-Bench | Vision & Documents · Document Parsing | headers-footers | 42.0% | #4 | 2026-04-20 | source ↗ |
| 13 | OCRBench | — | score | 880.00 | #11 | 2026-03-11 | source ↗ |
| 14 | olmOCR-Bench | — | pass-rate | 79.8% | #12 | 2026-03-11 | source ↗ |
| 15 | OmniDocBench | — | composite | 93.1% | #3 | 2026-03-11 | source ↗ |
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Strengths by area
Where Qianfan-OCR actually performs.
§ 06 · Sources & freshness
Where these numbers come from.
codesota-api
12
results
paperswithcode-public-api
3
results
3 of 15 rows marked verified. · first result 2026-03-11, latest 2026-04-20.