I'm an independent developer preparing to launch an AI chat app. I initially built it with Haiku because of the cost, and I'd like to compare it with GPT 5.6 Luna before release. However, my AWS account is new and doesn't currently have access to the GPT 5.6 model family, apparently because newer accounts may be restricted from using higher-cost models. Luna is now much cheaper than its original launch pricing and would be more affordable for my project, so I'd prefer to use it through AWS rather than another provider. I've already opened a support case, but I'm wondering whether moderators or an account team can help accelerate approval. Are there specific steps, justifications, or account requirements that might improve my chances?
3 Answers
Moderators don’t have a way to grant or accelerate model access. The support case is the appropriate route, although approval decisions are handled by the provider’s internal account and risk teams.
You could use the model through a direct provider or an aggregation service, but that would mean moving outside AWS. If keeping everything in AWS is important and you have credits available for initial testing, explain that clearly in the support case. Include your application’s use case, expected token volume, launch timeline, budget, and any safeguards you’ve implemented; concrete details may make the request easier to evaluate.
If you have an account team or an ISV representative, they may be able to open an internal request on your behalf. In my experience, they asked for details about the product, expected usage, business opportunities, and customers who would depend on the model. Larger established accounts often get access more easily, while a new basic account may not have that escalation path.
That makes sense. I’m an indie developer using a new basic account, so I don’t think I have access to an account team. I’ll include as much information as possible in the support request instead.
I’d prefer to keep the infrastructure under AWS and use my existing credits during testing because model usage will be my biggest early expense. I’ll make that context clear when I follow up.