DeepSeek V4 Pro (April preview)

DeepSeek V4 Pro, from the April 2026 preview, is the larger V4 model: an open-weight MoE with 49B of 1.6T active. It works in text, has a non-thinking mode and two thinking levels, targets agentic coding and agents, and is MIT-licensed.

Price per 1M tokens: $0.87 / $1.78input / output
Cached input: $0.08per 1M tokens
Context: 1M tokens384K max output

Providers

1 provider ·

Sort by

1M context · 384K max output

A request picks by price (:floor) or by speed (provider.sort, :nitro), or sets a floor on tokens per second. How routing works

Price history

No provider selected

Loading…

API

curl https://api.opencheap.si/v1/chat/completions \
  -H "Authorization: Bearer $OPENCHEAP_API_KEY" -H "Content-Type: application/json" \
  -d '{ "model": "opencheap/deepseek-v4-pro-20260423", "messages": [{ "role": "user", "content": "Hi" }] }'