Who leads the CC-OCR benchmark?

gemini-15-pro currently leads CC-OCR with a score of 83.25 on Multi-Scene F1.

What is the state-of-the-art score on CC-OCR?

The state-of-the-art result on CC-OCR is 83.25 (Multi-Scene F1), achieved by gemini-15-pro as of 2026.

How many models are tracked on CC-OCR?

Codesota tracks 5 models on CC-OCR across 4 metrics.

When was the CC-OCR leaderboard last updated?

The CC-OCR leaderboard on Codesota includes results through 2026.

Codesota · Benchmark · CC-OCRHome/Leaderboards/CC-OCR

South China University of Technology

CC-OCR.

Name: CC-OCR Benchmark Results
Creator: South China University of Technology
Published: 2026-01-01
License: https://creativecommons.org/licenses/by/4.0/

Multi-scene text reading, key information extraction, multilingual text, and document parsing benchmark.

Paper ↗Leaderboard ↓

§ 01 · Leaderboard

Results by metric.

Found a wrong score or missing run?

Use row edits to send a sourced correction into moderation.

Add / edit result ↗Report issue ↗

Multi-Scene F1

Multi Scene F1 is the reported evaluation metric for CC-OCR. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.

Higher is better

Trust tiers for Multi-Scene F1verifiedpapervendorcommunityunverified

Muted rows were not state of the art when published — an earlier or same-year result already scored better.

Rank	Model	Trust	Score	Year	Links	Fix
01	gemini-15-pro Non-API entry from src	unverified	83.25	2026	N/A	Looks wrong?
02	qwen2-vl-72b Non-API entry from src	unverified	77.95	2026	N/A	Looks wrong?
03	internvl2-76b Non-API entry from src	unverified	76.92	2026	N/A	Looks wrong?
04	gpt-4o Non-API entry from src	unverified	76.4	2026	N/A	Looks wrong?
05	claude-35-sonnet Non-API entry from src	unverified	72.87	2026	N/A	Looks wrong?

Multilingual F1

Multilingual F1 is the reported evaluation metric for CC-OCR. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.

Higher is better

Trust tiers for Multilingual F1verifiedpapervendorcommunityunverified

Muted rows were not state of the art when published — an earlier or same-year result already scored better.

Rank	Model	Trust	Score	Year	Links	Fix
01	gemini-15-pro Non-API entry from src	unverified	78.97	2026	N/A	Looks wrong?
02	gpt-4o Non-API entry from src	unverified	73.44	2026	N/A	Looks wrong?

KIE F1

Kie F1 is the reported evaluation metric for CC-OCR. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.

Higher is better

Trust tiers for KIE F1verifiedpapervendorcommunityunverified

Muted rows were not state of the art when published — an earlier or same-year result already scored better.

Rank	Model	Trust	Score	Year	Links	Fix
01	qwen2-vl-72b Non-API entry from src	unverified	71.76	2026	N/A	Looks wrong?
02	gemini-15-pro Non-API entry from src	unverified	67.28	2026	N/A	Looks wrong?
03	claude-35-sonnet Non-API entry from src	unverified	64.58	2026	N/A	Looks wrong?
04	gpt-4o Non-API entry from src	unverified	63.45	2026	N/A	Looks wrong?

Document Parsing

Document Parsing is the reported evaluation metric for CC-OCR. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.

Higher is better

Trust tiers for Document Parsingverifiedpapervendorcommunityunverified

Muted rows were not state of the art when published — an earlier or same-year result already scored better.

Rank	Model	Trust	Score	Year	Links	Fix
01	gemini-15-pro Non-API entry from src	unverified	62.37	2026	N/A	Looks wrong?

§ 04 · Submit a result

Add to the leaderboard.

Submit a Result

← Back to Leaderboards