PyyanThe AI Index

Pyyan / Compare / Qwen3-235B-A22B vs DeepSeek V4-Pro

Qwen3-235B-A22B vs DeepSeek V4-Pro

2 of 5

Open-Weight Models · verified 13 Sept 2026

Qwen3-235B-A22BAlibabacurrent
DeepSeek V4-ProDeepSeekcurrent
3 slots left
SpecificationQwen3-235B-A22BDeepSeek V4-Pro
SummaryCheapest capable model, most permissive licence.Best all-round open model of 2026.
LicenceApache-2.0MIT
Parameters235B MoE—
CategoryOpen-Weight ModelsOpen-Weight Models
OfficialAlibaba ↗DeepSeek ↗

Highlighted rows are where these differ.

Qwen3-235B-A22B

  • No licence conditions whatsoever
  • 22B active of 235B total

Best for commercial self-hosting.

Full spec sheet →

DeepSeek V4-Pro

  • Tops open leaderboards on agentic coding and reasoning
  • MIT licence — no conditions
  • Pioneered aggressive cache-hit pricing at ~$0.07/M
  • DeepSeek routes all V4-Pro requests to V4.1 Flash from 14 September 2026

Best for everything open.

Full spec sheet →