DeepSeek: DeepSeek-V3.2 (Non-thinking Mode)
Model Overview
| Property | Value |
|---|---|
| Model ID | deepseek/deepseek-chat |
| Name | DeepSeek-V3.2 (Non-thinking Mode) |
| Type | Chat/General Purpose |
Description
Main chat model with JSON output and tool calling capabilities. Optimized for conversational AI, general-purpose tasks, and integration with external tools and APIs.
Description
Model Overview
| Property | Value |
|---|---|
| Model ID | deepseek/deepseek-chat |
| Name | DeepSeek-V3. It features advanced reasoning capabilities for complex problem-solving tasks. The model is optimized for code generation and programming tasks. |
Specifications
| Spec | Value |
|---|---|
| Context Window | 128,000 tokens |
| Input Modalities | Text |
| Output Modalities | Text |
Capabilities
- JSON Output: Yes
- Tool Calling: Yes
- Reasoning/Thinking: No
- Code Execution: Yes
- Web Browsing: No
Best For
- Chat applications and conversational interfaces
- Tool integration and API orchestration
- JSON-structured data processing
- Business automation and workflow management
- Cost-sensitive applications with repeated queries
- Real-time customer support and chatbots
- General-purpose question answering
Pricing (Per 1M Tokens)
| Scenario | Input | Output |
|---|---|---|
| Cache Miss | $0.28 | $0.42 |
| Cache Hit | $0.028 | $0.42 |
Cache Mechanism
- Prompt caching reduces input cost by 90% (from $0.28 to $0.028 per 1M tokens)
- Repeated or similar content automatically matches against the cache
- First request pays full input price, subsequent requests benefit from the dramatically reduced cache hit rate
- Ideal for applications with repeated prompts, long-context systems, or batch processing
Special Features
- Advanced tool calling with structured outputs
- JSON mode for programmatic response formatting
- High-quality conversational performance
- Efficient cache mechanism for cost reduction
- Excellent for business logic and automation workflows
Integration
All models available via:
- DeepSeek API: https://api.deepseek.com
- OpenAI-compatible endpoints
- Standard /v1/chat/completions format
Example request:
curl -X POST https://api.langmart.ai/v1/chat/completions \
-H "Authorization: Bearer $DEEPSEEK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-chat",
"messages": [{"role": "user", "content": "Hello, how can you help me?"}],
"temperature": 0.7
}'
Model Family
DeepSeek-V3.2 represents the latest generation of DeepSeek models, offering significant improvements in reasoning, instruction-following, and multilingual capabilities compared to previous versions.