Missing model even with configured Api Key

[p44]

Hello all,

I configured API keys in env file: first for OpenAi, second for Anthropic, and third for Perplexity.

However, Perplexity doesn't appear among the models...

I also tried to configure the template in librechat.yaml, as per the guide, but nothing to do...

Any ideas?

[james]

Hello @p44
Did you restart the app after editing the env file?

[p44]

@james Yes, I followed all instructions... OpenAi and Anthropic they work fine, Perplexity no...

[james]

Hello @p44 just making sure.
Thanks for reporting.
I will look into it.

[p44]

@james Thanks James, waiting for your update

[james]

Hello @p44
Looking into it, but Perplexity does require payment setup for the API, so I am unable to generate API credentials for testing.

If you have some API credentials, would you be willing to provide them to me via a DM in the forum for testing?

[p44]

@james thanks a lot for your patience and your efforts. I just sent a DM with Perplexity Api Key.

[james]

Hello @p44

Thanks for providing the API Token.

@p44 said in Missing model even with configured Api Key:

I configured API keys in env file: first for OpenAi, second for Anthropic, and third for Perplexity.

This took me longer then expected.
It seems the variable you mentioned https://docs.openwebui.com/getting-started/env-configuration/#perplexity_api_key does not do what you think it does.

See: https://docs.openwebui.com/features/plugin/functions/pipe#extending-the-proxy-pipe

You can modify this proxy Pipe to support additional service providers like Anthropic, Perplexity, and more by adjusting the API endpoints, headers, and logic within the pipes and pipe functions.

So Perplexity is not supported in the same way OpenAI etc. is.
Adding the API URL and token via the connections tab did also not solve this.

After some more research, I found for Perplexity a custom Function is needed.
A custom what?

After even more search I found this: https://openwebui.com/f/neuromechanist/perplexity_pipe which I installed after a signup.
But you could also copy the code and add this function by hand.

After that I simply had to paste your API token:

Now I am able to select the modells:

and I asked it What is Cloudron, and should I use it to start a small business?

---

Cloudron is a server management platform designed to simplify self-hosting web applications like WordPress, Nextcloud, GitLab, and over 100 others. It automates deployment, updates, security, and maintenance, making it accessible for users without deep technical expertise[1][5].

Key Features for Small Businesses

• Automated Deployment:
  One-click installation for apps, including database provisioning, DNS configuration, and SSL certificates[1][5].
• Maintenance & Security:
  Automatic updates (including security patches within 24 hours), per-app backups, and offsite storage options (e.g., Amazon S3)[2][3].
  Built-in security measures like HTTPS, HSTS, rate limiting, and app isolation[3][5].
• Scalability:
  Immutable infrastructure allows easy migration of apps between servers[5].
  Supports user management with role-based access controls[1][3].

Pros vs. Cons for Small Businesses

Advantages	Limitations
Reduces IT overhead with automated updates and backups[1][2]	Limited free customer support; paid support may add costs[1]
Affordable starting price ($30/month) with no vendor lock-in[1][5]	Requires basic technical knowledge for troubleshooting[2]
Pre-configured apps streamline testing and deployment[4]	Limited customization for advanced server configurations[2]

Should You Use It?

Cloudron is ideal for small businesses prioritizing ease of use and security over deep customization. It’s particularly suited for:

• Tech-savvy teams needing rapid app deployment.
• Startups testing tools like GitLab, Rocket.Chat, or Nextcloud[1][4].
• Businesses with limited IT budgets, as it minimizes manual maintenance[2][5].

However, consider alternatives if you require 24/7 support or highly customized infrastructure. For most small businesses, Cloudron’s balance of automation and affordability makes it a strong contender[1][5].

References:
[1] https://www.softwareadvice.com/server-management/cloudron-profile/
[2] https://bitburners.com/a-brief-introduction-to-cloudron/
[3] https://forum.cloudron.io/topic/3849/how-does-cloudron-work-what-does-it-do-etc
[4] https://forum.cloudron.io/topic/2196/who-is-building-running-a-business-on-top-of-cloudron-what-s-your-model
[5] https://www.cloudron.io

---

I love that the AI used this Forum as a source. (:

[p44]

@james thanks for your efforts... it seems you worked on OpenWeb UI and not on LibreChat... right?

[james]

Hello @p44
Yes, I did. . .
Sorry for that, I will look into LibreChat also.

[james]

Second try.
I got it working in LibreChat much faster.
I added this to /app/data/env:

PERPLEXITY_API_KEY="YOUR-API-TOKEN-GOES-HERE"

And after reading https://www.librechat.ai/docs/configuration/librechat_yaml/ai_endpoints/perplexity and adding this config to the /app/data/librechat.yaml I found multiple issues with this sample config.
First there were some error about something being wrong in this yaml and after fixing that, the logs complained that:

Error 400 Invalid model 'llama-3-sonar-small-32k-chat'. Permitted models can be found in the documentation at https://docs.perplexity.ai/guides/model-cards.

So I fixed and tweaked this config to:

version: 1.2.4

endpoints:
  custom:
    - name: "Perplexity"
      apiKey: "${PERPLEXITY_API_KEY}"
      baseURL: "https://api.perplexity.ai/"
      models:
        default:
          - "sonar-deep-research"
          - "sonar-reasoning-pro"
          - "sonar-reasoning"
          - "sonar-pro"
          - "sonar"
          - "r1-1776"
        fetch: false
      titleConvo: true
      titleModel: "sonar"
      summarize: false
      summaryModel: "sonar"
      forcePrompt: false
      dropParams:
        - "stop"
        - "frequency_penalty"
      modelDisplayLabel: "Perplexity"

Which is working when using sonar, sonar-pro, sonar-reasoning, sonar-reasoning-pro but for some reason sonar-deep-research returns nothing.

I hope this helps and solves your issue.

[p44]

@james This solution works. Thanks a lot