DeepSeek OCR Models by DeepSeek
2 models2025–2026Up to 8K ctxFrom $0.03/1M input
About
DeepSeek OCR is DeepSeek's family of specialized vision-language models for optical character recognition and document parsing. DeepSeek-OCR introduces Contexts Optical Compression (arXiv 2510.18234) while DeepSeek-OCR-2 introduces Visual Causal Flow for improved document understanding (arXiv 2601.20552).
Current Variants
Use-when guidance is derived from seed capabilities, context, release, and replacement fields.
2 in view
DeepSeek OCR 2Current
Use when the workload needs vision, 8K context, and multimodal inputs.
2026-01vision8K contextmultimodal inputs
DeepSeek OCRCurrent
Use when the workload needs vision, 8K context, and multimodal inputs.
2025-10vision8K contextmultimodal inputs
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| DeepSeek OCR 2 | Use when the workload needs vision, 8K context, and multimodal inputs. | 2026-01 | vision8K contextmultimodal inputs | Current |
| DeepSeek OCR | Use when the workload needs vision, 8K context, and multimodal inputs. | 2025-10 | vision8K contextmultimodal inputs | Current |
Release Timeline
2 release groups2026-01
1 current
DeepSeek OCR 2
Currentvision8K contextmultimodal inputs
2025-10
1 current
DeepSeek OCR
Currentvision8K contextmultimodal inputs
Specifications(2 models)
| Model | Released | Context | Vision | Multimodal |
|---|---|---|---|---|
| DeepSeek OCR 2 | 2026-01 | 8K | Yes | Yes |
| DeepSeek OCR | 2025-10 | 8K | Yes | Yes |
Available From(1 provider)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| DeepSeek OCR | Novita AI | $0.03 | $0.03 | Serverless |
| DeepSeek OCR 2 | Novita AI | $0.03 | $0.03 | Serverless |
Frequently Asked Questions
- What is DeepSeek OCR used for?
- DeepSeek OCR is used for vision and vision and multimodal work. The family description and listed model capabilities point to those workloads as the best fit.
- How does DeepSeek OCR compare to Janus?
- DeepSeek OCR by DeepSeek is strongest where you need vision, while Janus by DeepSeek is the closest related family to check for image generation. DeepSeek OCR has 2 listed variants and reaches up to 8K context, so compare the specs and pricing tables before choosing a production model.
- Which DeepSeek OCR model should I use?
- For the lowest listed input price, start with DeepSeek OCR through Novita AI at $0.03/1M input tokens. For the most capable/latest local choice, evaluate DeepSeek OCR 2 with 8K context and multimodal inputs.
