Start here: how to use Agnes AI Discussions

[agnesai-admin]

Welcome to Agnes AI Discussions.

Use this space for open-ended questions, product ideas, roadmap feedback, integration patterns, and community examples around Agnes AI models and APIs.

The Agnes AI team will review and process discussion requests within 24 hours.

Please use GitHub Issues instead when you have a reproducible API integration problem, such as authentication errors, rate limits, request failures, video polling issues, or documentation gaps that require maintainer action.

Before posting, please do not include API keys, bearer tokens, private account data, customer content, or full private logs. If you need to share a request example, remove secrets and reduce it to a minimal sanitized example.

Helpful discussion topics include:

• Which model should I use for a workflow?
• How should I integrate Agnes AI with an agent or coding client?
• What examples or compatibility guides would help new developers?
• What product or API improvements would make Agnes AI easier to use?
• How are other teams using Agnes AI in production workflows?

Useful links:

• Developer docs: https://agnes-ai.com/doc/overview
• Model catalog: https://github.com/AgnesAI-SG/Agnes-AI/blob/main/MODEL_CATALOG.md
• Support guide: https://github.com/AgnesAI-SG/Agnes-AI/blob/main/SUPPORT.md
• Troubleshooting: https://github.com/AgnesAI-SG/Agnes-AI/blob/main/docs/TROUBLESHOOTING.md

[richardchen874-sys]

The Start here: how to use Agnes AI Discussions should be logged as a run-level decision, because model choice only matters if the task succeeds inside the latency and cost budget.

A compact provider matrix could track streaming, tool calls, structured output, usage reporting, timeout behavior, and fallback safety. That makes routing decisions easier to explain later.

This is close to what I am experimenting with: official Chinese models behind an OpenAI-compatible multi-model layer, with an emphasis on predictable usage and routing behavior. Which provider difference causes the most friction here: tools, streaming, usage reporting, or fallback behavior?