Does the new pricing and context length update affect legacy GPT‑3.5 models?

Below is some clarification regarding your questions:

Does the Price Change Also Apply to Older GPT‑3.5 Models?

No, the pricing changes noted on the pricing page apply to the new models (such as the updated GPT‑3.5‑turbo). The older GPT‑3.5 models have their own pricing details that are not overridden by the new models’ pricing. For example, in the deprecation documentation you can find tables that list shutdown dates, token costs, and recommended replacements for older models like gpt‑3.5‑turbo‑0613 and gpt‑3.5‑turbo‑16k‑0613.

Where to Find Specific Pricing Information for Older GPT‑3.5 Models

The detailed pricing for the older models is provided in the deprecation sections of the documentation. In those tables you’ll see information such as:

• Shutdown dates
• Deprecated model prices per 1M input/output tokens
• The recommended replacement model to transition to

You can find these details on the deprecation pages (for example, on https://platform.openai.com/docs/deprecations).

Does the Change in Context Length for GPT‑3.5 Also Affect the Older Models?

The update (released on November 6, 2023) introduces a new version of GPT‑3.5‑turbo that now comes with a 16k context by default. This change applies to the newer version of GPT‑3.5‑turbo. The older versions (such as gpt‑3.5‑turbo‑0613 and gpt‑3.5‑turbo‑16k‑0613) are being phased out – their existing capabilities, including context length, remain as they were until they are eventually deprecated. In other words, the new default context length is specific to the updated model, while the older models retain their previous features until they are replaced.

Always refer to the deprecation documentation for the most detailed transition plans and pricing specifics.