GLM 5.3 and DeepSeek V4.1 Flash resold at 60% off: the economics of open-source giveaways

lxfater · x · 2026-11-22

OpenCode now sells Zhipu's GLM 5.3 and DeepSeek V4.1 Flash at 60% below official pricing: $0.56 vs $1.40 per million input tokens for GLM 5.3, and $0.12 for DeepSeek V4.1 Flash — undercutting even DeepSeek's off-peak $0.15. Since both models are open-weight, resellers pay no training costs and just rent GPUs. The author's math for Chinese labs: (1) GPUs — they're compute-constrained, so third parties absorb serving traffic while official GPUs go to next-gen training; (2) customers — overseas paying users are locked in by OpenAI and Anthropic, so free distribution is the only way to get global coding tools and data-sensitive enterprises to integrate, monetized later via enterprise services. Users win with cheaper AI, but recouping training costs remains an open question.

Original post →

More from Venture

Venture channel →