What's this about?
OpenAI says subscription access to GPT-6 Astra and GPT-6.1 Sol now runs around 50% faster by default across its products and compatible partner tools, without user configuration.
What exactly is output speed?
It is usually measured in tokens per second: how quickly the model emits text after it starts answering. It is different from the delay before the first token and from total task time spent using tools.
What's the community saying?
18,244 likes, 887 reposts and 2,304 replies.
- Users welcomed a speed gain that does not consume the expensive fast-mode multiplier.
- Third-party coverage through Sign in with ChatGPT was an important part of the value.
- People still wanted clearer limits because faster generation can exhaust a quota sooner.