DeepSeek: DeepSeek V4 Pro
DeepSeek V4 Pro is an AI model with a vast context length of 1048576 tokens and supports both text input modalities and reasoning capabilities. It can be useful for tasks that require deep understanding and complex interactions, including coding and agentic decision-making. The model also supports tools integration, enhancing its versatility. Considering its benchmark blended score of 90.5 across four independent tests, DeepSeek V4 Pro is a robust choice for businesses and developers who need reliable performance in text-based applications. However, with pricing at $1.168 per million input tokens and $2.336 per million output tokens, it might be more costly compared to other models. For organizations prioritizing high-quality outcomes in coding and agentic tasks, the cost is justified by its strong performance metrics.
Benchmark results
Independent, published benchmarks. Blended score 90.5 across 4 benchmarks, last refreshed 2026-08-13. How scoring works →
| Benchmark | Measures | Score |
|---|---|---|
| AI Index | broad capability composite | 87.4 |
| AI Index Coding | software engineering tasks | 100.0 |
| AI Index Agentic | multi-step tool-using tasks | 81.8 |
| EQ-Bench | emotional understanding in dialogue | 77.6 |
- Model ID
- deepseek/deepseek-v4-pro
- Vendor
- deepseek
- Released
- April 2026
- Tokenizer
- DeepSeek
- Input Modalities
- text
- Output Modalities
- text
- Max Output
- 393,216 tokens
- Tool Calling
- ✓ supported
- Structured Output
- ✓ supported
- Reasoning Mode
- ✓ supported
- Vision
- text only
- Audio
- no
- Moderated
- no
What it costs in practice
Computed from the current $1.17/M input and $2.34/M output rates. Run your own numbers →
| Job | Tokens | Cost |
|---|---|---|
| Summarize a 50-page report | 30k in / 1.5k out | $0.04 |
| Classify 1,000 customer emails | 500k in / 50k out | $0.70 |
| A month of a busy support chatbot | 5M in / 2M out | $10.51 |
Price & spec history
Tracked daily by PicksByModel since 2026-07-17.
| Date | Input /M | Output /M | Context |
|---|---|---|---|
| 2026-08-13 | $1.17 | $2.34 | 1,048,576 |
| 2026-08-12 | $0.63 | $1.26 | 1,048,576 |
| 2026-08-11 | $0.69 | $1.37 | 1,048,576 |
| 2026-07-17 | $0.43 | $0.87 | 1,048,576 |
Category rankings
Where DeepSeek: DeepSeek V4 Pro places across the 53 categories it ranks in. How we rank →
| # | Category | Score |
|---|---|---|
| #7 | Code DocumentationCode · of 25 ranked | 148 |
| #7 | Regex WritingCode · of 25 ranked | 138 |
| #7 | Language TranslationWriting · of 25 ranked | 138 |
| #7 | Tutoring for KidsPersonal · of 25 ranked | 138 |
| #8 | SQL GenerationCode · of 25 ranked | 176 |
| #9 | Unit Test GenerationCode · of 25 ranked | 158 |
| #9 | CI/CD PipelinesCode · of 25 ranked | 148 |
| #9 | ETL ScriptingData · of 25 ranked | 160 |
| #9 | Medical Note SummarizationProfessional · of 25 ranked | 160 |
| #10 | Code ReviewCode · of 25 ranked | 175 |
| #10 | Code RefactoringCode · of 25 ranked | 173 |
| #10 | Bug FixingCode · of 25 ranked | 189 |
| #10 | CSV / Spreadsheet CleanupData · of 25 ranked | 157 |
| #10 | Long-Document SummarizationWriting · of 25 ranked | 168 |
| #10 | Blog Post WritingWriting · of 25 ranked | 144 |
| #10 | Creative WritingWriting · of 25 ranked | 165 |
| #10 | Screenwriting & DialogWriting · of 25 ranked | 165 |
| #10 | Transcript CleanupWriting · of 25 ranked | 145 |
| #10 | Academic WritingWriting · of 25 ranked | 165 |
| #10 | Legal DraftingProfessional · of 25 ranked | 165 |
| #10 | Legal ResearchProfessional · of 25 ranked | 177 |
| #10 | Contract ReviewProfessional · of 25 ranked | 187 |
| #10 | Scientific ResearchProfessional · of 25 ranked | 177 |
| #10 | Physics TutoringEducation · of 25 ranked | 153 |
| #10 | History TutoringEducation · of 25 ranked | 151 |
| #10 | Essay GradingEducation · of 25 ranked | 175 |
| #10 | RFP ResponseBusiness · of 25 ranked | 182 |
| #10 | OKRs & Strategic PlanningBusiness · of 25 ranked | 165 |
| #10 | Agent WorkflowsAgents · of 25 ranked | 201 |
| #10 | RAG PipelinesAgents · of 25 ranked | 173 |
| #10 | Coding AgentsAgents · of 25 ranked | 201 |
| #10 | Character RoleplayPersonal · of 25 ranked | 156 |
| #10 | Fiction CollaboratorPersonal · of 25 ranked | 177 |
| #10 | Trip PlanningPersonal · of 25 ranked | 144 |
| #10 | Math ProofsResearch · of 25 ranked | 165 |
| #10 | Scientific CodingResearch · of 25 ranked | 189 |
| #10 | Literature ReviewResearch · of 25 ranked | 177 |
| #11 | Long-Context Q&AAgents · of 25 ranked | 163 |
| #12 | Marketing CopyWriting · of 25 ranked | 132 |
| #12 | Standardized Test PrepEducation · of 25 ranked | 132 |
| #12 | Sales / Cold EmailBusiness · of 25 ranked | 123 |
| #12 | Job Application DraftingBusiness · of 25 ranked | 123 |
| #12 | Journaling HelperPersonal · of 25 ranked | 123 |
| #12 | Recipe GenerationPersonal · of 25 ranked | 123 |
| #13 | Math TutoringEducation · of 25 ranked | 151 |
| #14 | Meeting NotesBusiness · of 25 ranked | 154 |
| #14 | Fitness CoachingPersonal · of 25 ranked | 136 |
| #14 | Experiment DesignResearch · of 25 ranked | 161 |
| #15 | Data AnalysisData · of 25 ranked | 175 |
| #15 | Financial AnalysisProfessional · of 25 ranked | 163 |
| #21 | Resume WritingBusiness · of 25 ranked | 118 |
| #22 | SOP / Process DocsBusiness · of 25 ranked | 137 |
| #22 | Function / Tool CallingAgents · of 25 ranked | 153 |
Similar models
DeepSeek: DeepSeek V4 Pro 0813
DeepSeek: DeepSeek V4 Flash 0423
DeepSeek: DeepSeek V4 Flash 0731
DeepSeek: DeepSeek V3.2
DeepSeek: DeepSeek V3.2 Exp
DeepSeek: DeepSeek V3.1 Terminus
Quick answers
- How much does DeepSeek: DeepSeek V4 Pro cost?
- $1.17 per million input tokens and $2.34 per million output tokens.
- What is DeepSeek: DeepSeek V4 Pro's context window?
- 1,048,576 tokens, roughly 1,572 pages of text in a single request.
- Does DeepSeek: DeepSeek V4 Pro support tool calling?
- It supports tool calling, structured output, a reasoning mode.
- Can DeepSeek: DeepSeek V4 Pro process images?
- No, it is text-only on the input side.