DeepSeek V4 Pro as Your Coding Copilot
June 8, 2026
DeepSeek V4 Pro benchmarks competitively with GPT-4o on coding tasks while costing 89% less. Use it with Continue.dev, Aider, or Cursor via NexAPI.
Continue.dev Setup
Add to your Continue config:
provider: openai
model: deepseek-v4-pro
apiBase: https://nex-api.tech/v1
model: deepseek-v4-pro
apiBase: https://nex-api.tech/v1
Aider Setup
export OPENAI_API_BASE=https://nex-api.tech/v1
aider --model openai/deepseek-v4-pro
aider --model openai/deepseek-v4-pro
Cursor Setup
Settings, Models, Add OpenAI Compatible model with base URL nex-api.tech/v1
Benchmarks
| Benchmark | DeepSeek | GPT-4o |
|---|---|---|
| HumanEval | 92.1% | 90.2% |
| MBPP | 88.4% | 86.7% |
| MultiPL-E JS | 85.2% | 83.1% |