@molly0xFFF Seems like the email deliverability issue has been resolved (at least for me) but if you're still running into issues, one of the companies from my YC batch might be worth checking out: @resend (cc @zenorocha)
TIL that @AnthropicAI does not return an error if the tokens in your prompt + max_tokens_to_sample > 100K, which means that you can effectively emulate @OpenAI's option to leave max_tokens as null by setting max_tokens_to_sample = 100K. Neat (but also confusing API semantics).