DeepSeek V3 Models by DeepSeek
5 models2024–2025Up to 160k ctxFrom $0.1/1M input
Last refreshed 2026-05-22. Next refresh: weekly.
Details
ResearcherDeepSeek
LicenseMITOSI-approved
Commercial useCommercial use: permitted
Models5
Released2024–2025
Max context160k
Capabilities
Vision1 of 5 models
Multimodal1 of 5 models
JSON / Tool use1 of 5 models
Structured Outputs3 of 5 models
Code Execution2 of 5 models
Links
WebsiteAbout
DeepSeek V3 is a family of 5 AI models by DeepSeek, released between 2024 and 2025.
Decision facts
- Best fit
- vision and multimodal workJSON / Tool usestructured outputs
- Capability starting point
- DeepSeek V3.1 with 64k context and structured outputs and multimodal inputs
- Lowest tracked input
- DeepSeek V3 · $0.1/1M · Bitdeer AI
- Closest related family
- Janus
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
5 in view
DeepSeek V3.2Current
Use when the workload needs 160k context, 671B parameters, and structured outputs.
2025-12160k context671B parametersstructured outputs
DeepSeek V3.1Current
Use when the workload needs 64k context, structured outputs, and code execution.
2025-0864k contextstructured outputscode execution
DeepSeek V3 0324Current
Use when the workload needs 160k context and 671B parameters.
2025-03160k context671B parameters
DeepSeek V3Current
Use when the workload needs 64k context, 671B parameters, and JSON / Tool use.
2024-1264k context671B parametersJSON / Tool use
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| DeepSeek V3.2 | Use when the workload needs 160k context, 671B parameters, and structured outputs. | 2025-12 | 160k context671B parametersstructured outputs | Current |
| DeepSeek V3.1 | Use when the workload needs 64k context, structured outputs, and code execution. | 2025-08 | 64k contextstructured outputscode execution | Current |
| DeepSeek V3 0324 | Use when the workload needs 160k context and 671B parameters. | 2025-03 | 160k context671B parameters | Current |
| DeepSeek V3 Base | Use when the workload needs 128k context. | 2024-12 | 128k context | Current |
| DeepSeek V3 | Use when the workload needs 64k context, 671B parameters, and JSON / Tool use. | 2024-12 | 64k context671B parametersJSON / Tool use | Current |
Release Timeline
4 release groups2025-12
1 current
DeepSeek V3.2
Current160k context671B parametersstructured outputs
2025-08
1 current
DeepSeek V3.1
Current64k contextstructured outputscode execution
2025-03
1 current
DeepSeek V3 0324
Current160k context671B parameters
2024-12
2 current
DeepSeek V3
Current64k context671B parametersJSON / Tool use
DeepSeek V3 Base
Current128k context
Specifications(5 models)
| Model | Released | Context | Parameters | Vision | Multimodal | JSON / Tool use | Structured Outputs | Code Exec |
|---|---|---|---|---|---|---|---|---|
| DeepSeek V3.2 | 2025-12 | 160k | 671B | No | No | No | Yes | Yes |
| DeepSeek V3.1 | 2025-08 | 64k | 671B total, 37B active (MoE) | Yes | Yes | No | Yes | Yes |
| DeepSeek V3 0324 | 2025-03 | 160k | 671B | No | No | No | No | No |
| DeepSeek V3 Base | 2024-12 | 128k | 671B total, 37B active (MoE) | No | No | No | No | No |
| DeepSeek V3 | 2024-12 | 64k | 671B | No | No | Yes | Yes | No |
Available From(14 providers)
Pricing
Popular comparisons in this family
Comparisons
- GPT-4o Mini (07-18) vs DeepSeek V3
- GPT-4o (08-06) vs DeepSeek V3
- Claude 3.5 Sonnet vs DeepSeek V3
- Gemini 2.5 Flash vs DeepSeek V3
- DeepSeek V3 vs Llama 4 Maverick 17B Instruct FP8
- DeepSeek V3 vs DeepSeek R1
- DeepSeek V3.1 vs DeepSeek V4 Pro
- DeepSeek V3.1 vs GPT-5.5
Related Model Families
Models(5)
DeepSeek V3.2
2025-12160k671B7 providers
Open Source
DeepSeek V3.1
2025-0864k671B total, 37B active (MoE)8 providers
MultimodalOpen Source
DeepSeek V3 0324
2025-03160k671B3 providers
Open Source
DeepSeek V3 Base
2024-12128k671B total, 37B active (MoE)
Open Source
DeepSeek V3
2024-1264k671B13 providers
Open Source