Who leads the ThaiOCRBench benchmark?

claude-sonnet-4 currently leads ThaiOCRBench with a score of 0.84 on TED Score.

What is the state-of-the-art score on ThaiOCRBench?

The state-of-the-art result on ThaiOCRBench is 0.84 (TED Score), achieved by claude-sonnet-4 as of 2025.

Codesota tracks 10 models on ThaiOCRBench.

The ThaiOCRBench leaderboard on Codesota includes results through 2025.

Codesota · Benchmark · ThaiOCRBenchHome/Leaderboards/Vision & Documents/Document OCR/ThaiOCRBench

SCB 10X

Name: ThaiOCRBench Benchmark Results
Creator: SCB 10X
Published: 2025-01-01
License: https://creativecommons.org/licenses/by/4.0/

2,808 Thai text samples across 13 tasks. Tests Thai script structural understanding.

§ 01 · Leaderboard

Found a wrong score or missing run?

Use row edits to send a sourced correction into moderation.

Tree Edit Distance score for structural understanding

Higher is better

Trust tiers for TED Scoreverifiedpapervendorcommunityunverified

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	claude-sonnet-4 Thai OCR - Structural Understanding TED Score. 2,808 samples, 13 tasks	paper	0.84	2025	Source ↗	Looks wrong?
02	Claude Sonnet 4 Thai OCR - Structural Understanding TED Score. 2,808 samples, 13 tasks	unverified	0.84	2025	Source ↗	Looks wrong?
03	Gemini 2.5 Pro	unverified	0.77	2025	Source ↗	Looks wrong?
04	gemini-25-pro	paper	0.77	2025	Source ↗	Looks wrong?
05	Qwen2.5-VL 32B	unverified	0.77	2025	Source ↗	Looks wrong?
06	qwen25-vl-32b	paper	0.77	2025	Source ↗	Looks wrong?
07	internvl3-14b	paper	0.76	2025	Source ↗	Looks wrong?
08	InternVL3 14B	unverified	0.76	2025	Source ↗	Looks wrong?
09	Qwen2.5-VL 72B	unverified	0.72	2025	Source ↗	Looks wrong?
10	qwen25-vl-72b	paper	0.72	2025	Source ↗	Looks wrong?

Lineage

Predecessors (1)

This benchmark (1)

active2025-06

ThaiOCRBench

None yet — this is the current frontier.

§ 04 · Submit a result