Curious if anyone has gotten this to work with ollama locally?

[image] hendersj: FinalDestination::SSRFDetector::DisallowedIpError (FinalDestination: all resolved IPs were disallowed) lib/final_destination/ssrf_detector.rb:105:in 'lookup_and_filter_ips' lib/final_destination/http' That is our default SSRF protection kicking in. You can add exceptions to it via the DISCOURSE_ALLOWED_INTERNAL_HOSTS ENV var. It triggers when Discourse is trying to reach an IP considered internal.

I am also trying to get this to work with ollama, it seems that the API that is used from the Discourse AI plugin is not compatible with Ollama and any settings I change do not change the API request. This is what works with ollama: curl http://192.168.1.2:11434/api/generate -d '{ "model": "llama3.2", "prompt": "Why is the sky blue?" }' And above seems to be the only type of request the plugin is attempting. i.e. curl http://192.168.1.2:11434/ \ -X POST \ -H 'Content-Type: application/json' \ -d '{"inputs":" [INST] What is your favourite condiment? [/INST] Well, Im quite partial to a good squeeze of fresh lemon juice. It adds just the right amount of zesty flavour to whatever Im cooking up in the kitchen! [INST] Do you have mayonnaise recipes? [/INST]","parameters":{"max_new_tokens":500, "temperature":0.5,"top_p": 0.9}}' The AI Chat bot supposedly supports ollama but I am not able to get a response out of that either. So yeah anyone who has gotten this to work with ollama please post what settings you used!

Apparently I spoke too soon, if I do not have any settings in the Discourse AI plugin set and I follow the RAG instructions in this post I do get responses in the AI chat bot from Ollama!

It works just fine with ollama, we have multiple people on the team using it.

Cool. Is there a guide somewhere that we can follow on how to implement it? I’m more interested in the summarization and other functions than the chatbot. ETA: This configuration doesn’t work: [The image shows a configuration screen for an AI model named "olllama" hosted by OpenAI, with fields filled in for model ID, provider, URL, and API key, as well as options for tokenizing prompts and enabling visualization. (Captioned by AI)] (I redacted the IP address of my host here). What happens, is I get an Internal Server Error when I click “Run Test”. At /logs, I see: NameError (undefined local variable or method `tokenizer' for an instance of DiscourseAi::Completions::Dialects::ChatGpt) app/controllers/application_controller.rb:427:in `block in with_resolved_locale' app/controllers/application_controller.rb:427:in `with_resolved_locale' lib/middleware/omniauth_bypass_middleware.rb:35:in `call' lib/content_security_policy/middleware.rb:12:in `call' lib/middleware/anonymous_cache.rb:409:in `call' lib/middleware/csp_script_nonce_injector.rb:12:in `call' config/initializers/008-rack-cors.rb:14:in `call' config/initializers/100-quiet_logger.rb:20:in `call' config/initializers/100-silence_logger.rb:29:in `call' lib/middleware/enforce_hostname.rb:24:in `call' lib/middleware/processing_request.rb:12:in `call' lib/middleware/request_tracker.rb:385:in `call' This happens regardless of the tokenizer I select. I’m testing this on 3.5.0beta1-dev (c1ee4e120e).

Set the URL to http://localhost:11434/v1/chat/completions

[image] Falco: http://localhost:11434/v1/chat/completions The test still fails with the internal server error, same log message as well - a NameError saying that ‘tokenizer’ is an undefined local variable.

Try saving the model as is, refreshing the page and the coming back via the edit LLM button and running a new test.

Same result - I’ve cycled through that a few times, but just tried again. Did the install the usual way - added it to my app.yml file and ran ./launcher rebuild app The line I added was: hooks: after_code: - exec: cd: $home/plugins cmd: - sudo -E -u discourse git clone https://github.com/discourse/discourse-ai.git (I’ve removed the other plugins from the list but left the yml components in so you can see where this is in the yml file.)

I will get my local ollama setup back tomorrow and report here.

I did: ollama serve ollama run phi4 # test curl http://localhost:11434/v1/chat/completions \ -H "Content-Type: application/json" \ -d '{ "model": "phi4", "messages": [ { "role": "system", "content": "You are a helpful assistant." }, { "role": "user", "content": "How much is 1+1?" } ] }' -s | jq .choices[].message.content "The sum of 1 + 1 is 2. \n\nIn basic arithmetic, when you add one unit to another single unit, the total number becomes two units. This operation is part of fundamental addition principles used in mathematics. Is there anything else I can help with regarding math or any other topic?" [The image is a screenshot of a configuration interface for a software model named "ollama" with various settings, including name, model ID, provider, and service details, with successful submission indicated by a green "Success!" message. (Captioned by AI)] Tested it on Helper [The image shows a suggestion to replace "JavaScript" with the correct term "javascript" in a comment discussing adding JavaScript code to specific topics for ads. (Captioned by AI)]

Getting discourse ai to work with ollama locally

Product Support

Topic		Replies	Views
Discourse AI with local ollama Internal Server Error Bug ai	1	272	December 27, 2025
Self-Hosting an OpenSource LLM for DiscourseAI Self-Hosting ai	12	4608	July 6, 2026
Local Ollama is not working with the Plugin Product Support ai	3	446	October 15, 2025
Internal Server Error 500- Manual configuration [ Discourse AI] Product Support ai	7	514	September 5, 2025
Inquiry About AI Plugin Options Product Support ai	6	405	November 24, 2025

Getting discourse ai to work with ollama locally

Related topics