DeepSeek V3.2 Exp

DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism designed to improve training and inference efficiency in long-context scenarios while maintaining output quality. Users can control the reasoning behaviour with the `reasoning` `enabled` boolean. Learn more in our docs The model was trained under conditions aligned with V3.1-Terminus to enable direct comparison. Benchmarking shows performance roughly on par with V3.1 across reasoning, coding, and agentic tool-use tasks, with minor tradeoffs and gains depending on the domain. This release focuses on validating architectural optimizations for extended context lengths rather than advancing raw task accuracy, making it primarily a research-oriented model for exploring efficient transformer designs.

ChatDeepSeek164k tokens$0.2943 / $0.4469 · 1M

deepseek/deepseek-v3.2-exp

Contexto
164k tokens
Máx. completion
65,536
Tools
Sí
JSON
Sí
Lanzamiento
2025-09-29

Llámalo desde Geek Hub

El mismo SDK de OpenAI. Cambia el base URL y el id del modelo.

curl https://api.geekhub.mx/v1/chat/completions \
  -H "Authorization: Bearer $GEEKHUB_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v3.2-exp",
    "messages": [{"role": "user", "content": "Hola"}]
  }'
Consigue tu API key

Preguntas frecuentes

¿Qué es DeepSeek V3.2 Exp?
DeepSeek-V3.2-Exp is an experimental large language model released by DeepSeek as an intermediate step between V3.1 and future architectures. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse attention mechanism designed to improve training and inference efficiency in long-context scenarios while maintaining output quality. Users can control the reasoning behaviour with the `reasoning` `enabled` boolean. Learn more in our docs The model was trained under conditions aligned with V3.1-Terminus to enable direct comparison. Benchmarking shows performance roughly on par with V3.1 across reasoning, coding, and agentic tool-use tasks, with minor tradeoffs and gains depending on the domain. This release focuses on validating architectural optimizations for extended context lengths rather than advancing raw task accuracy, making it primarily a research-oriented model for exploring efficient transformer designs. DeepSeek V3.2 Exp corre en el API de Geek Hub (compatible con OpenAI). Id: deepseek/deepseek-v3.2-exp.
¿DeepSeek V3.2 Exp es gratis?
No. El input cuesta $0.2943 / 1M tokens y el output $0.4469 / 1M tokens en Geek Hub (markup incluido).
¿Cuál es el contexto de DeepSeek V3.2 Exp?
DeepSeek V3.2 Exp tiene una ventana de 164k tokens. Soporta hasta 65,536 tokens de completion.
¿DeepSeek V3.2 Exp soporta tool calling y structured outputs?
DeepSeek V3.2 Exp acepta tools y tool_choice para function calling. También soporta structured outputs con un JSON schema en response_format.
¿Cuándo se lanzó DeepSeek V3.2 Exp?
DeepSeek V3.2 Exp se lanzó el 29 de septiembre de 2025.

Más modelos de DeepSeek