Why doesn’t OpenAI offer DeepSeek or GLM as a cheaper option in codex?
retained text
The funny part is that I asked ChatGPT about this, and one of its explanations was basically that offering these models could take usage away from OpenAI’s own models. That made me laugh. I’m asking for a cheaper, smarter middle-tier option to replace Luna and Terra, which I barely use and the answer is essentially, “But then you might actually use it.” Would anyone else want this? What do you think is the biggest obstacle: cost, integration, or OpenAI’s business incentive to keep people using its own models? (I think there may be some misunderstanding. I’m specifically suggesting that OpenAI add open source models hosted on its own servers to the /model menu in Codex for subscribers, with usage counting toward the weekly allowance.)
document fetched 2026-10-25 00:21 · canonical https://reddit.com/r/ChatGPT/comments/1xdx22v/why_doesnt_openai_offer_deepseek_or_glm_as_a