Budibase/budibase · error · HTTPError

LLM not available

Error message

LLM not available

What it means

TableGeneration.init validates that an LLM configuration (aiModel) was resolved before constructing the generator. A missing/empty aiModel means there is no LLM available to perform AI table generation, so an HTTPError 422 is thrown early rather than failing mid-request.

Source

Thrown at packages/pro/src/ai/generators/tableGeneration.ts:44

    tables: Record<string, TableSchemaFromAI>
  ) => Promise<void>
}

type ProgressCallback = (message: string) => void

export class TableGeneration {
  private aiModel: LLMResponse

  private delegates: Delegates

  private constructor(aiModel: LLMResponse, delegates: Delegates) {
    this.aiModel = aiModel
    this.delegates = delegates
  }

  static async init(aiModel: LLMResponse, delegates: Delegates) {
    if (!aiModel) {
      throw new HTTPError("LLM not available", 422)
    }
    return new TableGeneration(aiModel, delegates)
  }

  async generate(
    userPrompt: string,
    userId: string,
    onProgress?: ProgressCallback
  ) {
    return tracer.trace("tableGeneration.generate", async span => {
      const tables = await this.generateTables(userPrompt, userId, onProgress)
      span.addTags({ table_count: tables.length })
      return tables
    })
  }

  private llmFunctions = {
    getTableStructure: async (

View on GitHub (pinned to a81a902e9a)

Solutions

1. Configure an LLM for the instance (AI settings / provider API keys) and select a model.
2. Ensure the AI proxy (e.g. LiteLLM at localhost:4000 in dev) is running and reachable.
3. Verify the license includes the AI feature if it is license-gated.
4. Guard the UI so the AI table generation option is hidden when no model is available, instead of calling the endpoint.

Example fix

// before
const tables = await generateTables({ prompt })
// after
const model = await getAvailableLLM()
if (!model) {
  toast.error("No AI model configured for this instance")
  return
}
const tables = await generateTables({ prompt })

const model = await getAvailableLLM()
if (!model) throw new Error("Configure an AI model before using AI table generation")

function hasAIModel(model: LLMResponse | null | undefined): model is LLMResponse {
  return model != null && typeof model === "object"
}

try {
  return await generateTables(req)
} catch (e) {
  if (e.status === 422 && e.message === "LLM not available") {
    promptUserToConfigureAI()
  } else throw e
}

When it happens

Trigger: Calling the AI table generation endpoint (POST /api/ai/tables) on an instance where no LLM is configured, the AI provider config failed to load, or the resolved aiModel is null/undefined (no model selected or AI feature not enabled/licensed).

Common situations: Self-hosted Budibase without AI/LLM environment configuration, missing provider API keys, license without AI feature, LiteLLM/AI proxy not running or unreachable so the model lookup returns nothing.

Related errors

• Error generating tables
• AI message content must be a string
• No response found
• AI user message must be a string
• AI system message must be a string

AI-assisted analysis of Budibase/budibase@a81a902e9a (2026-08-29). Data as JSON: /api/errors/56d38198ddf774ab. Report an issue: GitHub.