DeepSeek V4-Pro
DeepSeek's flagship open MoE for advanced reasoning, coding, and agentic work — 1.6T total parameters, 49B active, with a 1M-token context window. Built for the hardest tasks in the V4 lineup.
Architecture
MoE
1.6T total / 49B active
Context Window
1M
Tokens
Intelligence
50
AA Index v4.0
License
Open
Open Weights
Frontier V4 Flagship with 1M Context
DeepSeek V4-Pro is DeepSeek's flagship open MoE model for advanced reasoning, coding, and agentic work. With 1.6T total parameters, 49B active per token, and a 1M-token context window, it's designed for the hardest tasks in the V4 lineup.
V4-Pro shines on complex problem solving, knowledge-intensive workflows, and high-stakes coding or research use cases that demand frontier-level capability.
Frontier V4 Flagship
Built for the hardest reasoning & coding work
Advanced Reasoning
Frontier-level multi-step reasoning for the hardest problems in the V4 lineup.
High-Stakes Coding
Production-grade coding, refactoring, and full-repo generation across complex stacks.
Knowledge-Intensive Workflows
1M-token context handles entire research corpora, contracts, and codebases in a single call.
Agentic Research
Built for long-horizon agents: planning, tool use, and iterative experimentation at scale.
Frontier reasoning & coding performance
Artificial Analysis Intelligence Index v4.0 scores. V4-Pro is the flagship of the V4 family with frontier-level performance across reasoning, coding, and knowledge benchmarks.
Intelligence Index
Better than 94% of models
GPQA Diamond
Better than 95% of models
τ²-Bench Telecom
Better than 96% of models
| Category | Benchmark | Score |
|---|---|---|
| Reasoning | GPQA Diamond | 91 |
| Reasoning | Humanity's Last Exam | 34 |
| Reasoning | AA-LCR | 65 |
| Reasoning | CritPt | 10 |
| Coding | SciCode | 47 |
| Coding | Terminal-Bench Hard | 42 |
| Agentic | τ²-Bench Telecom | 95 |
| Agentic | IFBench | 71 |
| Agentic | GDPval-AA | 53 |
| Knowledge | AA-Omniscience Accuracy | 42 |
| Knowledge | AA-Omniscience Non-Hallucination | 13 |
Metrics sourced from Artificial Analysis and DeepSeek's published evaluations.
Flexible Pricing Tiers
Choose the optimal balance of speed and cost for your workflow. Prices are per 1M tokens.
| Tier | Input / 1M tokens | Output / 1M tokens |
|---|---|---|
| Overnight (24H) | $1.05 | $2.20 |
| Async | $1.31 | $2.75 |
| Realtime | $1.74 | $3.48 |
Context window natively supported up to 1M tokens.
Start Building in Minutes
DeepSeek V4-Pro is accessible via OpenAI-compatible endpoints.
from openai import OpenAI
client = OpenAI(
api_key="your-api-key-here",
base_url="https://api.doubleword.ai/v1"
)
# Frontier reasoning with DeepSeek V4-Pro (1M context)
response = client.chat.completions.create(
model="deepseek-ai/DeepSeek-V4-Pro",
messages=[
{"role": "user", "content": "Solve this multi-step research problem and show your reasoning."}
],
)
print(response.choices[0].message.content)💡 Pro Tip
V4-Pro is built for the hardest tasks — reach for it on frontier reasoning, complex coding, and knowledge-intensive research workflows. For lighter everyday work, V4-Flash offers similar capability at a fraction of the cost.