All tasks / Comparisons
PaddleOCR-VL-1.6 vs TeleOCR
Compared only on public benchmark results, across the 1 task where both are listed: they share a tier in 1.
PaddleOCR-VL-1.6
Baidu's open-weights 0.9B vision-language model for page-level document parsing of text, tables and formulas.
- Vendor
- Baidu (PaddlePaddle)
- Weights
- Open: run it yourself
- Released
- 28 May 2026
- GPU memory to run locally
- ~4 GB
- License
- Apache-2.0
- Last verified
- 18 Oct 2026
TeleOCR
Open-weights ~1.2B vision-language model for parsing digital and camera-captured documents; formerly NaviDC-OCR.
- Vendor
- XingChen-AGI
- Weights
- Open: run it yourself
- Released
- 17 Aug 2026
- GPU memory to run locally
- ~5 GB
- License
- Apache-2.0
- Last verified
- 18 Oct 2026
Task by task
| Task | PaddleOCR-VL-1.6 | TeleOCR | Higher tier |
|---|---|---|---|
| Documents & OCR Models | S96 on OmniDocBench v1.6 (end-to-end, v1.6_full) · Local | S97 on OmniDocBench v1.6 (end-to-end, v1.6_full) · Local | Same tier |
A tier spans a range of scores, and small gaps on a leaderboard are often within its margin of error, so only a different tier counts as ahead. How we rank.
Same benchmark, both results
| Benchmark | PaddleOCR-VL-1.6 | TeleOCR | Snapshot |
|---|---|---|---|
| OmniDocBench v1.6 (end-to-end, v1.6_full) Overall page-parsing score: ((1 − text edit distance) × 100 + table TEDS + formula CDM) / 3, higher is better | 96 | 97 | 11 Sep 2026 |
| ParseBench (models) Overall: mean of five rule-based dimension scores (tables, charts, content faithfulness, semantic formatting, visual grounding) on ~2,000 human-verified enterprise pages, higher is better | 67.4% | 58.7% | 9 Oct 2026 |