Skip to main content

TeleAISpecialist modelOpen weights

TeleOCR

PopularityRank 5 of 19 specialist models

38out of 100new calculation

usage data only, range 25 to 52

Summary

TeleOCR is a lightweight vision-language model for document parsing. It handles digital and camera-captured documents and targets tables, formulas, layout, and content. The checkpoint has approximately 1.2 billion parameters and is released for local execution.

The weights were first released as NaviDC-OCR on August 17, 2026. The project renamed the model to TeleOCR on September 10. It remains one Apache-2.0 checkpoint rather than two separate models.

All benchmark results

Every published result with its source, test setup and comparison models.

Show all 39 published benchmark results

Page 1 of 4

Measurements without matching peer values

These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.

Dr.DocBench overall (Show measurement, test conditions, and source)
Source value
67.96
Metric
unknown
Unit
No unit provided
Category
document-ocr
Direction
Higher is better
Source type
official-leaderboard
Evaluator
Dr.DocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Document-parsing challenge; historical row label NaviDC-OCR, mapped by the model card to TeleOCR native weights.
Context
Source-defined overall metric. The canonical profile retains the older navidc-ocr ID while displaying TeleOCR.
Source Dr.DocBench
Dr.DocBench text edit (Show measurement, test conditions, and source)
Source value
0.1903
Metric
edit-distance
Unit
edit-distance
Category
document-ocr
Direction
Lower is better
Source type
official-leaderboard
Evaluator
Dr.DocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Document-parsing challenge; historical NaviDC-OCR native-weight row.
Source Dr.DocBench
Dr.DocBench formula CDM (Show measurement, test conditions, and source)
Source value
0.02
Metric
unknown
Unit
No unit provided
Category
document-ocr
Direction
Higher is better
Source type
official-leaderboard
Evaluator
Dr.DocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Document-parsing challenge; historical NaviDC-OCR native-weight row.
Context
Source-defined metric scale; the row's positive direction is retained.
Source Dr.DocBench
Dr.DocBench table TEDS (Show measurement, test conditions, and source)
Source value
64.97
Metric
unknown
Unit
No unit provided
Category
document-ocr
Direction
Higher is better
Source type
official-leaderboard
Evaluator
Dr.DocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Document-parsing challenge; historical NaviDC-OCR native-weight row.
Context
Source-defined metric scale.
Source Dr.DocBench
Dr.DocBench order edit (Show measurement, test conditions, and source)
Source value
0.398
Metric
edit-distance
Unit
edit-distance
Category
document-ocr
Direction
Lower is better
Source type
official-leaderboard
Evaluator
Dr.DocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Document-parsing challenge; historical NaviDC-OCR native-weight row.
Source Dr.DocBench
OmniDocBench v1.6 overall (Show measurement, test conditions, and source)
Source value
96.87
Metric
unknown
Unit
No unit provided
Category
document-ocr
Direction
Higher is better
Source type
official-leaderboard
Evaluator
OmniDocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Specialized VLM table; TeleOCR 1.2B row.
Context
Source-defined overall metric.
Source OmniDocBench
OmniDocBench v1.6 text edit (Show measurement, test conditions, and source)
Source value
0.027
Metric
edit-distance
Unit
edit-distance
Category
document-ocr
Direction
Lower is better
Source type
official-leaderboard
Evaluator
OmniDocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Specialized VLM table; TeleOCR 1.2B row.
Source OmniDocBench
OmniDocBench v1.6 formula CDM (Show measurement, test conditions, and source)
Source value
96.36
Metric
unknown
Unit
No unit provided
Category
document-ocr
Direction
Higher is better
Source type
official-leaderboard
Evaluator
OmniDocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Specialized VLM table; TeleOCR 1.2B row.
Context
Source-defined metric scale.
Source OmniDocBench
OmniDocBench v1.6 table TEDS (Show measurement, test conditions, and source)
Source value
97.05
Metric
unknown
Unit
No unit provided
Category
document-ocr
Direction
Higher is better
Source type
official-leaderboard
Evaluator
OmniDocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Specialized VLM table; TeleOCR 1.2B row.
Context
Source-defined metric scale.
Source OmniDocBench
OmniDocBench v1.6 table TEDSS (Show measurement, test conditions, and source)
Source value
98.52
Metric
unknown
Unit
No unit provided
Category
document-ocr
Direction
Higher is better
Source type
official-leaderboard
Evaluator
OmniDocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Specialized VLM table; TeleOCR 1.2B row.
Context
Source-defined metric scale.
Source OmniDocBench
OmniDocBench v1.6 read-order edit (Show measurement, test conditions, and source)
Source value
0.122
Metric
edit-distance
Unit
edit-distance
Category
document-ocr
Direction
Lower is better
Source type
official-leaderboard
Evaluator
OmniDocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Specialized VLM table; TeleOCR 1.2B row.
Source OmniDocBench
Wild OmniDocBench overall (Show measurement, test conditions, and source)
Source value
88.53
Metric
unknown
Unit
No unit provided
Category
document-ocr
Direction
Higher is better
Source type
official-leaderboard
Evaluator
Wild OmniDocBench
Status
active
Retrieved at
2026-10-03
Methodology
Official benchmark leaderboard. Wild document benchmark; TeleOCR 1.2B row.
Context
Source-defined overall metric.
Source Wild OmniDocBench

Page 1 of 4

Popularity compared

How often the model is searched for, used and mentioned, compared with every model of its category.

Popularity index

0 to 100, higher is more popular

20 of 21 models

As of November 2026 · Source: gradually.ai popularity index

Over time

Weekly values from October 18, 2026 to November 20, 2026

TeleOCRKumo-Tabular 45Agora-2 33Qwen-Drive-1.0-4B 29AstaBrief-8B 29

Method changes in the timeline, a jump at these points is no market event:

  1. 1November 1, 2026: Value only with search or usage data, 30-day attention
  2. 2November 1, 2026: Method version 4 (common scale, shrinkage)
  3. 3November 9, 2026: New topics measurement V20 (daily quorum, short forms)

What is gaining popularity right now, across models, vendors, harnesses and AI subscriptions, is collected in All AI trends.

All specifications

Architecture

Model class
Vision-language model for document parsing
Source: Model class
Parameters
Approximately 1.2 billion
Source: Parameters

Watermarking and provenance

AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.

AI watermarking
Unknown
no source
Watermarking technique
Not known
no source

Licence and access

License
Apache-2.0
Source: License
Shutdown
None announced
no source

Further details

Current model ID
XingChen-AGI/TeleOCR
Source: Current model ID
Former model ID
NaviDC-OCR, renamed to TeleOCR on September 10, 2026
Source: Former model ID
Documents
Digital and camera-captured documents with tables and formulas
Source: Documents

Sources and data date

Show all 16 sources

Benchmark sources at a glance

Which independent evaluators tested this model. This is coverage, not a quality ranking.

No active result from the reviewed external evaluators can currently be attributed to this model. Other source evidence remains available below.

Without vendor claims and historical or disputed results; those stay labelled in the single values.

TypeEvidence and data date
Research statusResearch date October 4, 2026. 9 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Unresolved

remaining-ocr-subtasks

Note

PureDocBench reports Overall for Clean, Digital and Real documents and an Avg3 value for TeleOCR. Higher values are better. Separate FormulaCDM and TableTEDS component values are not shown in the main leaderboard. Sci-ImageMiner RMS, TEDS and weighted values remain unassigned because their direction is not established by the inspected sources.

Additional sourceTeleOCR model card (retrieved October 3, 2026; October 23, 2026) · TeleOCR model card · Editorial description reviewed October 3, 2026
Additional sourceTeleOCR technical report (retrieved October 3, 2026) · TeleOCR technical report · Editorial description reviewed October 3, 2026
Additional sourceDr.DocBench (retrieved October 3, 2026)
Additional sourceOmniDocBench (retrieved October 3, 2026)
Additional sourceWild OmniDocBench (retrieved October 3, 2026)
Additional sourcePureDocBench (retrieved October 3, 2026; October 18, 2026)
Additional sourceTeleOCR model card, ICDAR2026 Sci-ImageMiner table (retrieved October 18, 2026)
Additional sourceHugging Face TeleOCR evaluation-results commit 5e9317b (retrieved October 23, 2026)
Additional sourceModel metadata (retrieved October 3, 2026)
Additional sourceModel metadata (retrieved October 23, 2026)
Additional sourceOpenRouter: Popularity index (derived), CC BY 4.0
Additional sourceVercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional sourceOllama: cumulative downloads for the popularity index
Additional sourceReplicate: cumulative runs for the popularity index
Additional sourceGoogle Trending Now: limited feed selection for Germany and the US
Additional sourceCreative Commons Attribution 4.0