Budibase/budibase · error · HTTPError

Unsupported BBAI model: ${model}

Error message

Unsupported BBAI model: ${model}

What it means

Budibase AI v2 chat completions only accept model identifiers namespaced with the 'budibase/' prefix. If the request body's model field is set to anything else (OpenAI-style names, custom strings, empty-ish variants), the endpoint rejects it with HTTP 400 to prevent routing to unknown providers.

Source

Thrown at packages/server/src/api/controllers/ai/budibaseai-v2.ts:80

  }
}

export async function chatCompletionV2(ctx: Ctx<ChatCompletionRequestV2>) {
  const { messages, model, stream } = ctx.request.body

  if (!messages?.length) {
    ctx.throw(400, "Missing required field: messages")
  }

  if (env.SELF_HOSTED && !env.isDev()) {
    ctx.throw(500, "Budibase AI endpoints are not available in self-host")
  }

  if (!model) {
    ctx.throw(400, "Missing required field: model")
  }
  if (!model?.startsWith("budibase/")) {
    throw new HTTPError(`Unsupported BBAI model: ${model}`, 400)
  }

  const bbaiKey = bbai.getBBAIKey()

  await quotas.throwIfBudibaseAICreditsExceeded()

  const requestBody = {
    ...ctx.request.body,
    metadata: {
      ...ctx.request.body.metadata,
      tags: getBBAITags(),
    },
  }

  const upstreamResponse = await fetch(
    `${environment.LITELLM_URL.replace(/\/$/, "")}/chat/completions`,
    {
      method: "POST",

View on GitHub (pinned to a81a902e9a)

Solutions

1. Prefix the model with 'budibase/', e.g. model: 'budibase/gpt-4o-mini'
2. Check the deployed Budibase version for the exact list of supported budibase/ model ids
3. If you need a non-budibase provider, call that provider's API directly instead of the BBAI endpoint

Example fix

// before
{ "model": "gpt-4o-mini", "messages": [...] }
// after
{ "model": "budibase/gpt-4o-mini", "messages": [...] }

if (!model?.startsWith("budibase/")) {
  throw new Error(`Model must start with "budibase/", got: ${model}`)
}

const isBbaiModel = (m: unknown): m is `budibase/${string}` =>
  typeof m === "string" && m.startsWith("budibase/")

try {
  await bbaiChat({ model, messages })
} catch (e) {
  if (e.status === 400 && String(e.message).includes("Unsupported BBAI model")) {
    // correct the model id and retry once
  }
}

When it happens

Trigger: POST to the BBAI v2 chat completion endpoint with body.model such as 'gpt-4o', 'claude-3', or a typo'd id that does not start with 'budibase/'.

Common situations: Clients configured for the generic OpenAI-compatible API reused against Budibase AI; model lists copied from other providers; older SDK versions pre-dating the budibase/ prefix convention.

Related errors

• AI message content must be a string
• Tool name must be under 64 characters long
• Cannot create a query tool without a query ID
• Invalid bookmark query
• Invalid limit query

AI-assisted analysis of Budibase/budibase@a81a902e9a (2026-08-29). Data as JSON: /api/errors/9b1741b11b844cee. Report an issue: GitHub.