OpenAI opens the ChatGPT API at a tenth of the price

OpenAI opened API access to the model behind ChatGPT, branded gpt-3.5-turbo, at $0.002 per 1,000 tokens — roughly a tenth of the price of the company’s previous GPT-3.5-family models. The same announcement introduced a Whisper API for speech-to-text transcription, priced separately, giving developers hosted access to both OpenAI’s flagship chat model and its speech-recognition system through a single platform.

Until this release, building ChatGPT-like products meant working with OpenAI’s older completion-style models, priced and tuned differently from the chat interface millions of users had adopted since ChatGPT’s November launch. Opening a cheaper, chat-tuned model to developers collapsed that gap: companies could now build directly on the same underlying model driving the consumer product, at a fraction of the previous cost per token.

The price cut was aggressive enough that outside observers described it as OpenAI treating API revenue as a volume business rather than a premium one — a strategic choice that assumed usage would scale to offset thinner per-token margins, and one that put pressure on competitors’ pricing for comparable models. Snap, Instacart, Quizlet and Shopify were among the companies that announced gpt-3.5-turbo integrations around the launch.

The API opening, arriving two weeks before GPT-4’s release, triggered a wave of applications built on hosted language models rather than open-weight alternatives, and it is generally treated as the moment building a product on top of a frontier model’s API became a viable, inexpensive strategy for start-ups rather than a research-lab-only capability.