DeepSeek V4 Pro
Released 2026-04 · hybrid · 1.0M tokens · 9 benchmarks · Open weight
Editorial notes
Preview lanzado 24 abril 2026. 1.6T total / 49B activos. Hybrid Attention (CSA + HCA). 3 niveles de effort: Non-think / Think High / Think Max. 27% FLOPs y 10% KV-cache vs V3.2 en context 1M. Codeforces 3206. MIT license. Pricing $1.74/$3.48 per MTok (cache $0.015, -99%). Reemplaza deepseek-chat y deepseek-reasoner.
Spec sheet
- Empresa
- DeepSeek
- Pais
- CN
- Tipo
- hybrid
- Release
- 2026-04
- Context
- 1.0M tokens
- Params total
- 1.6T
- Params active (MoE)
- 49B
- Licencia
- mit
- Quants
- BF16, Q8_0, Q5_K_M, Q4_K_M, Q3_K_M
- Pricing (deepseek)
- $0.435/$0.87/M
- Slug
- deepseek-v4-pro
Quick install
2 toolsollama run deepseek-v4:proNota: Q4_K_M requiere multi-GPU/multi-node
vllm serve deepseek-ai/DeepSeek-V4-Pro --tensor-parallel-size 8Benchmarks (9)
Reasoning 3
Math 2
Coding 4
- 93.5LiveCodeBenchProblemas de coding contests en vivo de LeetCode/Codeforces.
- 80.6SWE-bench-VerifiedIssues reales de GitHub de 12 repos populares de Python.
- 76.8HumanEvalCorreccion funcional en 164 problemas de codigo Python.
- 55.4SWE-bench-ProVersion profesional de SWE-bench con issues mas complejos.
Cite this model
BibTeX · APA
BibTeX
@misc{frontier-deepseek-v4-pro,
title = {DeepSeek V4 Pro},
author = {{DeepSeek}},
year = {2026},
note = {Frontier Benchmarks AI atlas. Accessed 2026-05-08},
url = {https://frontierbenchmarks.com/models/deepseek-v4-pro}
} APA
DeepSeek (2026). DeepSeek V4 Pro [Large language model]. Frontier Benchmarks AI. Retrieved 2026-05-08, from https://frontierbenchmarks.com/models/deepseek-v4-pro
Citation refleja la pagina del atlas, no el paper original del modelo. Para el paper, ve a la seccion "Recursos" arriba.