DeepSeek has permanently reduced the price of DeepSeek V4 Pro by 75%
$0.003625 per million input tokens (with cache)
$0.435 per million input tokens
$0.87 per million output tokens
Cache is almost free
CodexSaver routes low-risk tasks from Codex to DeepSeek, keeping costly models for decisions—48% average savings across five tasks with ~6s latency.
MIMO V2.5 Pro beats DeepSeek V4 Pro with faster output and cleaner visuals, thanks to its multimodal roots. DeepSeek V4 Vision may close the gap.
Lift is a 9B open-source model for extracting structured data from documents via JSON Schema, reaching 90.2% accuracy. Install with pip install lift-pdf.