Show HN: Opper AI – Task-Completion API for LLMs(opper.ai)
opper.ai
Show HN: Opper AI – Task-Completion API for LLMs
https://opper.ai/
22 comments
Co-Founder here thanks for taking a look at Opper! I’m hanging around the thread all day, so feel free to ask anything, share feedback, or tell us where you’d like the product to go next
Do you have a demo video available for the latest features/use cases? We are building a digital end-of-life service platform with customized "Bereavement AI" where we have tried to switch between GPT-4o mini and Gemini 2.5 Flash for different types user questions. Would be interesting to give this a spin in our use case!
No up to date demo video unfortunately :(
Sounds like a great use case though!
Sounds like a great use case though!
What are your thoughts on longetivity, i.e. depending on your service over time?
Do you have any public commitments in that regard?
Being able to run your application forever is one of the strong benefits I see with local models, besides being able to feed it with sensitive data.
Do you have any public commitments in that regard?
Being able to run your application forever is one of the strong benefits I see with local models, besides being able to feed it with sensitive data.
We have been thinking a bit about this, and one option would be to have some form of locally hosted runner. You can optimize the task in the cloud and deploy it locally. Something like that. It is possible to plug in custom models so technically feasible.
Congrats, team! I've been following you for quite some time, happy to see you outta stealth and crushing it!
My only question is: What has been the most exciting learning experience you've had while building Opper so far?
My only question is: What has been the most exciting learning experience you've had while building Opper so far?
I think just how far you can go with examples has been an interesting learning! As these models have become smarter, they are also getting better at reasoning from examples and understanding intent. We will be publishing some research in the next few days!
Hey, I am the founder of TenderRender. We're looking into question specific in-context learning examples. So retrieving different examples based on the prompt. Is this possible?
Yes that's possible! You can populate examples of great outputs to task specific datasets and have those be automatically populated to the prompt. More info here: https://docs.opper.ai/capabilities/learning
Been using Opper for a while, and my absolute favorite way of building AI apps. Talk about DX and implementation speed!
Thanks for the shout out!
Long-time user of Opper! Can recommend! Makes building LLM apps so much faster.
Thanks happy to hear Niels!
Hey, other co-founder here. Happy to answer questions.
Really cool tool! Great job!
Awesome!
LLM features often fail in production due to fragile, model-specific prompt chains. That’s why we built a task-completion API to make model calls as reliable as any other API: you declare what you want, and it manages the interaction with 80+ proprietary and open-source models. If the output fails, Opper retries or falls back automatically. Successful completions can be saved as task specific dataset entries and serve as examples for future generations. When the completion passes (or retries are exhausted), you get structured output plus a pass/fail flag.
For example, gettested.io previously spent weeks rewriting prompts to generate consistent blood test summaries. With Opper, a single task definition now delivers reports in 62 countries without any edits.
Key features:
- Define tasks in JSON: inputs, outputs, success test
- Automatic prompt construction, retries and fallbacks
- Quality through task specific datasets and in-context learning
- Full observability (llm-as-a-judge, prompts, responses, tokens, costs)
- Free tier up to $5/mon. Utility plan from $5/month gets you high rate limits
We built Opper after our last startup, Unomaly (ML observability, acquired 2020). Our vision with Opper is to let developers work with LLMs the same way they write code and get reliable results.
The task-completion API is live today. We’d love for you to give it a try and would appreciate any feedback you have.
Thanks for checking it out!