OpenAI Watch posted an update
A tester comparing GPT-6.1 Sol in Codex and through the API says Ultrafast was marginally faster in Codex, within the variation of the test. In regular fast mode, the tester recorded 42 tokens per second in Codex against 88 through the API.
Why it mattersThat is one person’s comparison, not a controlled benchmark, and the post’s account of usage costs is cut off before it gives the full detail. Still, it is a useful reminder that a model’s speed can depend on where and how it is used. Which matters more in practice: headline speed, or predictable usage costs?
Discuss: When choosing an AI coding setup, would you prioritise measured speed in your own workflow or clearer, more predictable usage costs?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.