All the questions addressed by the API team during the December 17, 2024 AMA
FreeOpenAI API team answers developer questions from December 17, 2024 AMA
FreeFree tier
About All the questions addressed by the API team during the December 17, 2024 AMA
A community forum post summarizing the questions and answers from OpenAI's API team AMA held on December 17, 2024, as part of the Shipmas event. The post covers key topics such as structured outputs streaming bugs, the absence of a Sora API, the current status of the Assistants API and planned o1 integration, Whisper V3 updates, GPT-4o mini audio preview model, and input token limit discussions.
Key Features
Addresses structured outputs streaming newline bug with troubleshooting advice
Confirms no immediate plans for Sora API but invites feedback
Assistants API remains active; o1 integration planned for next year
No Whisper V3 update yet; recommends GPT-4o mini audio preview for better speech understanding and TTS control
Discusses input token limits; team acknowledges need for increased limits
Covers upcoming improvements for GPT-4o mini audio preview model
Pros & Cons
Pros
- Provides direct, official answers from the OpenAI API team
- Covers a wide range of developer concerns in a single post
- Offers practical advice for common API issues (e.g., schema simplification for streaming)
- Transparent about future plans and limitations
Cons
- Information is brief and may lack depth for complex technical issues
- Some answers provide no concrete timelines (e.g., Sora API, Whisper V3)
- Not a comprehensive reference; only covers questions asked during the AMA
Best For
Developers troubleshooting OpenAI API issues like structured outputs streamingUnderstanding OpenAI's roadmap for new models and API featuresPlanning integrations with Assistants API and future o1 supportExploring audio understanding and TTS capabilities with GPT-4o mini audio preview
FAQ
Is there a Sora API available?
OpenAI has no plans for a Sora API yet, but they welcome developer feedback on potential use cases.
Will the Assistants API support o1?
Yes, the team plans to bring o1 to the Assistants API next year, with more details to be shared soon.
Is Whisper V3 coming to the API?
No announcement yet regarding Whisper V3. For audio understanding and TTS, the GPT-4o mini audio preview model is recommended.
How can I fix the streaming structured outputs newline bug?
This can occur when the model outputs improbable tokens; try using a simpler schema or adjusting prompting to be clearer.
When will input token limits be increased?
The team acknowledges the need for higher input token limits and is working on it, but no specific timeline was given.