Scan to join WeChat groupDeepSeek Reasoner
DeepSeek Reasoner: a deepseek-thinking model from DeepSeek, ~1M context, knowledge cutoff 2025-09
DeepSeek Reasoner is a deepseek-thinking model from DeepSeek (~1M context, input around $0.14/1M tokens). Suitable for assistants, content generation, knowledge Q&A and business automation
starsCapabilities
codeFunction calling
paymentsContext and pricing
Context limit1,000,000
Max output384,000
Knowledge cutoff2025-09
Input price$0.14/ 1M tokens
Output price$0.28/ 1M tokens
Cached input price$0.0028/ 1M tokens
descriptionOverview
Overview
DeepSeek Reasoner is provided by DeepSeek, model ID deepseek-reasoner
Key specs
- Context: 1M tokens
- Max output: 384K tokens
- Knowledge cutoff: 2025-09
- Input price: $0.14/1M
- Output price: $0.28/1M
Best for
Consider DeepSeek Reasoner when comparing context length, pricing, multimodal support and relay availability
lightbulbUse cases
- Assistants and customer support
- Content generation and rewriting
- Knowledge Q&A and summarization
- Structured extraction
thumb_upStrengths
- Very long context (~1M), good for long documents and codebases
- Open weights, can be self-hosted
- High single-response output limit (~384K)
infoLimitations
- Knowledge cutoff 2025-09; newer facts need external retrieval
- Pricing and availability vary by upstream and relay; verify with official docs and tests