Skip to content

can't create a Conversation Search(Build-in RAG) model #450

Open

Description

@nexus-4

I can't create a model. I am following the documentation:

1 -> first step ok: {"message": "A collection with name conversation_store already exists."}%

2 step: {"message": "Gemini API error: "}%

and this is what I use to get the error(changed the API Key of course with the one I recently created in ai studio:

curl 'http://localhost:8108/conversations/models'
-X POST
-H 'Content-Type: application/json'
-H "X-TYPESENSE-API-KEY: xyz"
-d '{
"id": "conv-model-1",
"model_name": "google/gemini-2.0-flash",
"history_collection": "conversation_store",
"api_key": "AIaksdjfadsjbf",
"system_prompt": "You are an assistant for question-answering. You can only make conversations based on the provided context. If a response cannot be formed strictly using the provided context, politely say you do not have knowledge about that topic.",
"max_bytes": 16384
}'

{"message": "Gemini API error: "}%

Activity

  1. nexus-4 commented on Mar 12, 2026

    @nexus-4
    Author

    I also tested my api key before that using:
    curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=apikey" \ -H 'Content-Type: application/json' \ -X POST \ -d '{ "contents": [{ "parts":[{"text": "Hello!"}] }] }'

    response:
    curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=apikey" \ -H 'Content-Type: application/json' \ -X POST \ -d '{ "contents": [{ "parts":[{"text": "Hello!"}] }] }' { "candidates": [ { "content": { "parts": [ { "text": "Hi there! How can I help you today?\n" } ], "role": "model" }, "finishReason": "STOP", "avgLogprobs": -0.14047146927226672 } ], "usageMetadata": { "promptTokenCount": 2, "candidatesTokenCount": 11, "totalTokenCount": 13, "promptTokensDetails": [ { "modality": "TEXT", "tokenCount": 2 } ], "candidatesTokensDetails": [ { "modality": "TEXT", "tokenCount": 11 } ] }, "modelVersion": "gemini-2.0-flash", "responseId": "e8OyaZjzG7n-qtsPj8yf2QY" }

  2. tharropoulos commented on Mar 12, 2026

    @tharropoulos
    Collaborator

    Hey @nexus-4,
    This isn't related to this repository. I didn't get an error while trying to use this model myself, which version of Typesense are you on? The only issue I've faced is

    This model model_name is no longer available to new users. Please update your code to use a newer model for the latest features and improvements

    cc: @jasonbosco could you transfer this? I don't have access.

  3. nexus-4 commented on Mar 12, 2026

    @nexus-4
    Author

    I just noticed that, sorry.
    but, I installed via homebrew...

    curl http://localhost:8108/health
    {"ok":true}%

    curl -H 'X-TYPESENSE-API-KEY: xyz' http://localhost:8108/debug
    {"state":1,"version":"30.1"}%

    I just tested once more and got the same thing: {"message": "Gemini API timeout."}%

  4. tharropoulos commented on Mar 12, 2026

    @tharropoulos
    Collaborator

    This is a different error though, a timeout vs a generic "API error". Could you check it with another model?

  5. nexus-4 commented on Mar 13, 2026

    @nexus-4
    Author

    Some context:

    • MacOs;
    • version 30.1(homebrew);
    • Models tested: 2.0, 2.5-flash and 3-flash-preview

    First, I tried direct calling and this was one of the responses:

    curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-3-flash-preview:generateContent?key=apikey" \
    -H 'Content-Type: application/json' \
    -X POST \
    -d '{
      "contents": [{
        "parts":[{"text": "Olá! Você está funcionando?"}]
      }]
    }'
    {
      "candidates": [
        {
          "content": {
            "parts": [
              {
                "text": "Olá! Sim, estou funcionando perfeitamente. Sou uma inteligência artificial e estou pronto para ajudar.\n\nComo posso ser útil para você hoje? Você tem alguma dúvida, precisa de ajuda com algum texto ou quer conversar sobre algum assunto?",
                "thoughtSignature": "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"
              }
            ],
            "role": "model"
          },
          "finishReason": "STOP",
          "index": 0
        }
      ],
      "usageMetadata": {
        "promptTokenCount": 6,
        "candidatesTokenCount": 48,
        "totalTokenCount": 295,
        "promptTokensDetails": [
          {
            "modality": "TEXT",
            "tokenCount": 6
          }
        ],
        "thoughtsTokenCount": 241
      },
      "modelVersion": "gemini-3-flash-preview",
      "responseId": "5RC0aZCNJ6Plz7IPwOnE8QE"
    }
    

    Then, when I try to use any of the models above to create a model, an error:

    curl 'http://localhost:8108/conversations/models' \
      -X POST \
      -H 'Content-Type: application/json' \
      -H "X-TYPESENSE-API-KEY:xyz" \
      -d '{
            "id": "conv-model-1",
            "model_name": "google/gemini-2.5-flash",
            "history_collection": "conversation_store",
            "api_key": "apikey",
            "system_prompt": "You are an assistant for question-answering. You can only make conversations based on the provided context. If a response cannot be formed strictly using the provided context, politely say you do not have knowledge about that topic.",
            "max_bytes": 16384
          }'
    
    {"message": "Gemini API error: {"message": "Bad request."}"}%   
    

    Now, this is a different error, but there's always an error while trying to create a model.

    • {"message": "Gemini API error: "}%
    • {"message": "Gemini API timeout."}%
    • {"message": "Gemini API error: {"message": "Bad request."}"}%
  6. tharropoulos commented on Mar 13, 2026

    @tharropoulos
    Collaborator

    This for sure isn't reproducible on my end. I've tried it with several API keys, several models and all work, even on real queries against the Conversational Search feature.

    Tested it both under:

    • EndeavourOS Linux 6-19.7
    • MacOS Darwin 25.2.0

    On Docker using Typesense 30.1.

    ❯ curl 'http://localhost:8108/conversations/models' \
      -X POST \
      -H 'Content-Type: application/json' \
      -H "X-TYPESENSE-API-KEY: xyz" \
      -d '{
            "id": "model1",
            "model_name": "google/gemini-2.5-flash",
            "history_collection": "conversation_store",
            "api_key": "Salazar**********************************",
            "system_prompt": "You are an assistant for question-answering. You can only make conversations based on the provided context. If a response cannot be formed strictly using the provided context, politely say you do not have knowledge about that topic.",
            "max_bytes": 16384
          }'
    {"api_key":"Salazar**********************************","history_collection":"conversation_store","id":"model1","max_bytes":16384,"model_name":"google/gemini-2.5-flash","system_prompt":"You are an assistant for question-answering. You can only make conversations based on the provided context. If a response cannot be formed strictly using the provided context, politely say you do not have knowledge about that topic.","ttl":86400}%
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions