Deepseek-V4-Pro-Preview

Model IntroductionText

DeepSeek-V4-Pro-Preview is a 1.6-trillion-parameter Mixture-of-Experts model that delivers million-token context with state-of-the-art open-source performance.

Released on April 24, 2026, DeepSeek-V4-Pro-Preview is the flagship preview model in DeepSeek's V4 series[reference:0]. It features 1.6 trillion total parameters with 49 billion activated per token, built on a Mixture-of-Experts architecture that supports a context length of one million tokens. The model is designed to work with agent frameworks including Claude Code and OpenClaw, reflecting the industry shift toward models that can complete complex, multi-step tasks with less human input. Performance-wise, it rivals top closed-source systems in agentic coding, world knowledge, and competitive programming.

The model introduces a Hybrid Attention Architecture combining Compressed Sparse Attention (CSA) and Heavily Compressed Attention (HCA), dramatically improving long-context efficiency—requiring only 27% of single-token inference FLOPs and 10% of KV cache compared to DeepSeek-V3.2 in the 1M-token context setting. It incorporates Manifold-Constrained Hyper-Connections (mHC) to strengthen signal propagation across layers, and employs the Muon optimizer for faster convergence and greater training stability. Pre-trained on over 32T diverse and high-quality tokens, the model follows a two-stage post-training paradigm: independent cultivation of domain-specific experts, followed by unified consolidation via on-policy distillation. In maximum reasoning mode, DeepSeek-V4-Pro-Max achieves top-tier performance in coding benchmarks and significantly bridges the gap with leading closed-source models on reasoning and agentic tasks.

Benchmarks

Benchmark (Area)Score*
HLE (Reasoning)37.7 (Eval)
HLE (w/ Tools) (Reasoning)48.2 (Eval)
CritPt (Reasoning)12.9 (Eval)
AIME 2026 (Reasoning)94.6 (Eval)
HMMT Nov. 2025 (Reasoning)94.4 (Eval)
HMMT Feb. 2026 (Reasoning)95.2 (Eval)
IMOAnswerBench (Reasoning)89.8 (Eval)
GPQA-Diamond (Reasoning)90.1 (Eval)
SWE-bench Pro (Coding)55.4 (Eval)
NL2Repo (Coding)35.5 (Eval)
DeepSWE (Coding)8 (Eval)
ProgramBench (Coding)47.8 (Eval)
Terminal Bench 2.1 (Terminus-2) (Coding)64 (Eval)
Terminal Bench 2.1 (Best Reported Harness) (Coding)- (Eval)
FrontierSWE (Dominance) (Coding)29.0 (Eval)
PostTrainBench (Coding)- (Eval)
SWE-Marathon (Coding)- (Eval)
MCP-Atlas (Public Set) (Agentic)73.6 (Eval)
Tool-Decathlon (Agentic)52.8 (Eval)

API request example

Endpoint /v1/chat/completions. Replace <API_KEY> with a key from the console.

Standard OpenAI Chat Completions. Swap `model` for the target slug.

curl https://api.ravenrouter.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{
  "model": "Deepseek-V4-Pro-Preview",
  "messages": [
    {
      "role": "user",
      "content": "Hello! Introduce yourself in one sentence."
    }
  ],
  "max_tokens": 64
}'

More models: API Reference → Examples by model

Deepseek-V4-Pro-Preview API — $1.93/1M input, $3.86/1M output | RavenRouter AI