chat Text Deep reasoning

DeepSeek V4 Pro

Strong at complex, multi-step work

Trained at larger scale with deep reasoning (thinking) mode. Strong for multi-step work like refactoring and debugging, and for agent workflows that chain tool calls.

verified

Single price — no hidden costs

Parameters: 1.6T total · 49B active (MoE)
License: MIT (open-weight)
Context: 1M tokens
Max output: 384K tokens
Tool calling: Supported (multi-step agent chains)
Input: $0.435 / 1M tokens
Output: $0.87 / 1M tokens
Cache Read /M: $0.003625 / 1M tokens
30-day uptime: 99.9%
Time to first response: ~700ms
Throughput: ~60 tok/s

Measured by meshcode across our own infrastructure — redundancy means uptime holds even if one upstream has issues.

Use this model View API docs

Quality

Reasoning

Speed

Cost efficiency

Good fit for

Code refactoring Root-cause debugging Multi-tool agent work Large-scale design/architecture review

Code example

curl https://api.meshcode.ai/v1/chat/completions \
  -H "Authorization: Bearer $MESHCODE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-pro",
    "messages": [{ "role": "user", "content": "Hello!" }]
  }'

FAQ

Which company built this model? expand_more

DeepSeek develops and operates this model directly. meshcode contracts API access and passes calls straight through.

Are there any hidden costs? expand_more

No. The input/output prices shown here are the whole story — no subscription fee, no minimum commitment. You're only charged for what you use, deducted from your wallet balance.