Alibaba

Qwen2.5-VL 32B Instruct (DeepInfra)

Accuracy: 51.6%$0.0021/page

Qwen/Qwen2.5-VL-32B-Instruct — OCR-capable vision model.

Providers for Qwen2.5-VL 32B Instruct (DeepInfra)

OpenOCR routes requests to the best provider able to handle your file type and parameters, with fallbacks to maximize uptime.

DeepInfra

☁️ CloudLive

Latency

5184ms

Throughput

—

Price/page

$0.0021/page

Max pages/req

Formatspngjpgjpegwebpgif

Specifications

Benchmark score

51.6%benchmarked

Latency

5.2s

Languages

100+

Max Pages

Supported Formats

pngjpgjpegwebpgif

Capabilities

printed-texthandwritingmulti-languagelayout-analysisscene-text

Strengths & Weaknesses

Strengths

Data Privacy

No data trainingUS

Data retentionRetention unknown

Aggregator. OpenOCR routes only to paid tiers in prod; free training tiers are excluded. Per-tier privacy varies upstream.

Engine Comparison

Engine	Provider	Benchmark	Latency	Price/Page	Languages
Mistral OCR 4	Mistral AI	84.8%✓	2.0s	$0.005/page	0
Meta: Llama 4 Maverick	OpenRouter	71%✓	2.9s	$0.00036/page	100
Anthropic: Claude Opus 4.5	OpenRouter	67.5%✓	4.4s	$0.0085/page	100
Meta: Muse Spark 1.2	OpenRouter	65.5%✓	11.2s	$0.0095/page	0
Mistral: Mistral Medium 3	OpenRouter	64.9%✓	4.3s	$0.0008/page	100
OpenAI: GPT-5.6 Luna	OpenRouter	64.8%✓	5.4s	$0.0017/page	0
Mistral: Mistral Medium 3.1	OpenRouter	64.6%✓	3.1s	$0.00078/page	100
StepFun: Step 3.7 Flash	OpenRouter	63.6%✓	25.2s	$0.00064/page	0
Gemini 3 Flash Preview	Google Cloud	63.6%✓	7.0s	~$0.0008/page	100
Anthropic: Claude Opus 4.6	OpenRouter	63.4%✓	4.5s	$0.0087/page	100
Qwen2.5-VL 32B Instruct (DeepInfra)current	DeepInfra	51.6%✓	6.8s	$0.0021/page	100

Top 10 by benchmark score, plus this engine. Compare all 128 engines

This engine supports all languages via its underlying AI model. Pass any ISO 639-1 language code to optimize results.

100+ languages supported

Trust & Privacy