Need Advise - openai agents sdk vs langgraph agents and gpt -5

Hey guys - I think i’m tired and exhausted. Building a new app, usually use langchain but this time started with openai swarm first and then moved to agents -sdk and using gpt-5 models. It was confusing to use agents sdk responses but running into issues with tool calling with gpt-5 models, unpredicable responses sometimes and latency is god knows. Spending way too much time so swtiched over to langchain/langraph but then looks gpt-5 removed stop sequences that langchain/langraph adds when call llms

So back to openai sdk.

Now at a point, should I just stop using openai sdk’s and models and use anthropic / grok ?

If you haven’t already, you can also try using agents sdk with LiteLLM and other models.

This way you can easily switch back to openai models when things get better, and will also have alternatives in a situation you’ve hit a wall with a particular model or provider.

Thanks. I used LiteLLM earlier but need to check latest to see if i can i run multi-agents with that.