Skip to content

OpenRAG is a comprehensive, single package Retrieval-Augmented Generation platform built on Langflow, Docling, and Opensearch.

Python

fix: make onboarding model catalog discoverable (#2370)

Edwin Josefc489d846c82
diff --git a/frontend/app/onboarding/_components/advanced.test.tsx b/frontend/app/onboarding/_components/advanced.test.tsx
index 73af59d06..f587a57ca 100644
--- a/frontend/app/onboarding/_components/advanced.test.tsx
+++ b/frontend/app/onboarding/_components/advanced.test.tsx
@@ -4,6 +4,38 @@ import { renderWithProviders, screen, userEvent } from "@/test-utils/render";
import { AdvancedOnboarding } from "./advanced";
describe("AdvancedOnboarding custom models", () => {
+ it("previews five models, exposes the full list, and explains Azure search", async () => {
+ const user = userEvent.setup();
+ const models = [
+ ...Array.from({ length: 5 }, (_, index) => ({
+ value: `gpt-5-${index}`,
+ label: `gpt-5-${index}`,
+ })),
+ { value: "gpt-4.1", label: "gpt-4.1" },
+ ];
+
+ renderWithProviders(
+ <TooltipProvider>
+ <AdvancedOnboarding
+ languageModels={models}
+ languageModel="gpt-5-0"
+ setLanguageModel={vi.fn()}
+ searchPlaceholder="Search models or type Azure deployment name"
+ />
+ </TooltipProvider>,
+ );
+
+ await user.click(screen.getByTestId("language-model-selector"));
+ expect(
+ screen.getByPlaceholderText(
+ "Search models or type Azure deployment name",
+ ),
+ ).toBeInTheDocument();
+ expect(screen.queryByRole("option", { name: "gpt-4.1" })).toBeNull();
+ await user.click(screen.getByRole("option", { name: "Show all 6 models" }));
+ expect(screen.getByRole("option", { name: "gpt-4.1" })).toBeInTheDocument();
+ });
+
it("accepts a language model that is absent from provider inventory", async () => {
const setLanguageModel = vi.fn();
const user = userEvent.setup();
diff --git a/frontend/app/onboarding/_components/advanced.tsx b/frontend/app/onboarding/_components/advanced.tsx
index d87ee424e..a7cdabc7f 100644
--- a/frontend/app/onboarding/_components/advanced.tsx
+++ b/frontend/app/onboarding/_components/advanced.tsx
@@ -7,6 +7,7 @@ export function AdvancedOnboarding({
embeddingModels,
languageModel,
embeddingModel,
+ searchPlaceholder,
setLanguageModel,
setEmbeddingModel,
}: {
@@ -15,6 +16,7 @@ export function AdvancedOnboarding({
embeddingModels?: { value: string; label: string }[];
languageModel?: string;
embeddingModel?: string;
+ searchPlaceholder?: string;
setLanguageModel?: (model: string) => void;
setEmbeddingModel?: (model: string) => void;
}) {
@@ -39,6 +41,8 @@ export function AdvancedOnboarding({
<ModelSelector
options={embeddingModels}
custom
+ previewLimit={5}
+ searchPlaceholder={searchPlaceholder}
data-testid="embedding-model-selector"
icon={icon}
value={embeddingModel}
@@ -56,6 +60,8 @@ export function AdvancedOnboarding({
<ModelSelector
options={languageModels}
custom
+ previewLimit={5}
+ searchPlaceholder={searchPlaceholder}
data-testid="language-model-selector"
icon={icon}
value={languageModel}
diff --git a/frontend/app/onboarding/_components/generic-onboarding.test.tsx b/frontend/app/onboarding/_components/generic-onboarding.test.tsx
index 5d7934df3..8206ef7cf 100644
--- a/frontend/app/onboarding/_components/generic-onboarding.test.tsx
+++ b/frontend/app/onboarding/_components/generic-onboarding.test.tsx
@@ -1,91 +1,138 @@
-import { render, screen } from "@testing-library/react";
-import userEvent from "@testing-library/user-event";
-import { describe, expect, it, vi } from "vitest";
-import { TooltipProvider } from "@/components/ui/tooltip";
+import { HttpResponse, http } from "msw";
+import type { Dispatch, SetStateAction } from "react";
+import { describe, expect, it } from "vitest";
+import type { OnboardingVariables } from "@/app/api/mutations/useOnboardingMutation";
+import type { SavedProvidersSnapshot } from "@/components/models/catalog-models";
+import { renderWithProviders, screen, userEvent } from "@/test-utils/render";
import { GenericOnboarding } from "./generic-onboarding";
-vi.mock("@/app/api/queries/useGetModelsQuery", () => ({
- useGetModelCatalogQuery: vi.fn().mockReturnValue({
- data: {
- providers: [
+const catalog = {
+ providers: [
+ {
+ key: "azure",
+ name: "Azure OpenAI",
+ credential_fields: [],
+ models: [{ model: "gpt-5.5", capabilities: ["function_calling"] }],
+ embedding_models: [{ model: "text-embedding-3-small" }],
+ },
+ {
+ key: "openai_like",
+ name: "Other provider",
+ credential_fields: [
{
- key: "openai",
- models: [{ model: "gpt-4o" }, { model: "gpt-4o-mini" }],
- embedding_models: [
- { model: "text-embedding-3-small" },
- { model: "text-embedding-3-large" },
- ],
+ key: "api_base",
+ label: "API base",
+ required: false,
+ field_type: "text",
},
],
+ models: [{ model: "gpt-4.1", capabilities: ["function_calling"] }],
+ embedding_models: [
+ { model: "text-embedding-3-small" },
+ { model: "text-embedding-3-large" },
+ ],
},
- }),
-}));
-
-vi.mock("@/components/models/model-selector", () => ({
- ModelSelector: ({ value, onValueChange, options = [] }: any) => (
- <select
- data-testid="model-selector"
- value={value}
- onChange={(e) => onValueChange?.(e.target.value)}
- >
- {options.map((m: any) => (
- <option key={m.value} value={m.value}>
- {m.label}
- </option>
- ))}
- </select>
- ),
-}));
+ ],
+};
-describe("GenericOnboarding", () => {
- it("seeds credentials and synchronizes parent settings", async () => {
- const user = userEvent.setup();
- const mockSetSettings = vi.fn();
+function renderOnboarding(
+ provider: string,
+ isEmbedding = false,
+ savedProviders?: SavedProvidersSnapshot,
+) {
+ let settings: OnboardingVariables = {};
+ const setSettings: Dispatch<SetStateAction<OnboardingVariables>> = (next) => {
+ settings = typeof next === "function" ? next(settings) : next;
+ };
+ const user = userEvent.setup();
+ renderWithProviders(
+ <GenericOnboarding
+ provider={provider}
+ isEmbedding={isEmbedding}
+ setSettings={setSettings}
+ providers={savedProviders}
+ />,
+ {
+ providers: ["tooltip"],
+ handlers: [
+ http.get("/api/models/catalog", () => HttpResponse.json(catalog)),
+ ],
+ },
+ );
+ return { user, getSettings: () => settings };
+}
- render(
- <TooltipProvider>
- <GenericOnboarding
- provider="openai"
- isEmbedding={false}
- setSettings={mockSetSettings}
- providers={
- {
- custom: { openai: { credential_values: { api_key: "sk-test" } } },
- } as any
- }
- />
- </TooltipProvider>,
+describe("GenericOnboarding model selection", () => {
+ it.each([
+ {
+ kind: "language",
+ isEmbedding: false,
+ selector: "language-model-selector",
+ modelKey: "llm_model",
+ catalogModel: "gpt-5.5",
+ },
+ {
+ kind: "embedding",
+ isEmbedding: true,
+ selector: "embedding-model-selector",
+ modelKey: "embedding_model",
+ catalogModel: "text-embedding-3-small",
+ },
+ ] as const)("requires an explicit Azure $kind model choice", async (testCase) => {
+ const { user, getSettings } = renderOnboarding(
+ "azure",
+ testCase.isEmbedding,
);
+ const selector = screen.getByTestId(testCase.selector);
+ await user.click(selector);
+ const catalogOption = await screen.findByRole("option", {
+ name: testCase.catalogModel,
+ });
- expect(mockSetSettings).toHaveBeenCalled();
+ expect(selector).toHaveTextContent("Select model...");
+ expect(getSettings()[testCase.modelKey]).toBeFalsy();
- // Check input change
- const apiKeyInput = screen.getByLabelText(/API Key/i);
- await user.clear(apiKeyInput);
- await user.type(apiKeyInput, "sk-new-key");
+ await user.click(catalogOption);
+ expect(getSettings()[testCase.modelKey]).toBe(testCase.catalogModel);
+ });
- expect(mockSetSettings).toHaveBeenCalled();
+ it("keeps automatic defaults for other generic providers", async () => {
+ const { user, getSettings } = renderOnboarding("openai_like");
+ await user.click(screen.getByTestId("language-model-selector"));
+ await screen.findByRole("option", { name: "gpt-4.1" });
+ expect(getSettings().llm_model).toBe("gpt-4.1");
});
- it("handles embedding mode and model selection", async () => {
- const user = userEvent.setup();
- const mockSetSettings = vi.fn();
+ it("seeds saved credentials and synchronizes edits to parent settings", async () => {
+ const { user, getSettings } = renderOnboarding("openai_like", false, {
+ custom: {
+ openai_like: {
+ credential_values: { api_base: "https://saved.example" },
+ },
+ },
+ });
+ await user.click(screen.getByTestId("language-model-selector"));
+ await screen.findByRole("option", { name: "gpt-4.1" });
- render(
- <TooltipProvider>
- <GenericOnboarding
- provider="openai"
- isEmbedding={true}
- setSettings={mockSetSettings}
- providers={{ custom: {} } as any}
- />
- </TooltipProvider>,
+ const apiBase = screen.getByLabelText("API base");
+ expect(apiBase).toHaveValue("https://saved.example");
+ expect(getSettings().provider_credentials?.openai_like?.api_base).toBe(
+ "https://saved.example",
);
- expect(mockSetSettings).toHaveBeenCalled();
+ await user.clear(apiBase);
+ await user.type(apiBase, "https://new.example");
+ expect(getSettings().provider_credentials?.openai_like?.api_base).toBe(
+ "https://new.example",
+ );
+ });
- const selector = screen.getByTestId("model-selector");
- await user.selectOptions(selector, "text-embedding-3-large");
- expect(mockSetSettings).toHaveBeenCalled();
+ it("updates the selected embedding model for another generic provider", async () => {
+ const { user, getSettings } = renderOnboarding("openai_like", true);
+ await user.click(screen.getByTestId("embedding-model-selector"));
+ await user.click(
+ await screen.findByRole("option", { name: "text-embedding-3-large" }),
+ );
+ expect(getSettings().embedding_model).toBe("text-embedding-3-large");
});
});
diff --git a/frontend/app/onboarding/_components/generic-onboarding.tsx b/frontend/app/onboarding/_components/generic-onboarding.tsx
index 13d426550..435274177 100644
--- a/frontend/app/onboarding/_components/generic-onboarding.tsx
+++ b/frontend/app/onboarding/_components/generic-onboarding.tsx
@@ -10,6 +10,7 @@ import { useGetModelCatalogQuery } from "@/app/api/queries/useGetModelsQuery";
import { LabelInput } from "@/components/label-input";
import {
onboardingCredentialFields,
+ providerCatalogOptions,
type SavedProvidersSnapshot,
savedCredentialValuesForProvider,
savedSecretFieldsForProvider,
@@ -127,23 +128,22 @@ export function GenericOnboarding({
}));
};
- const catalogEntry = catalog?.providers?.find(
- (entry) => entry.key === provider,
+ const models = useMemo(
+ () =>
+ providerCatalogOptions(
+ catalog,
+ provider,
+ isEmbedding ? "embedding" : "language",
+ ),
+ [catalog, provider, isEmbedding],
);
- const models = useMemo(() => {
- const entries = isEmbedding
- ? (catalogEntry?.embedding_models ?? [])
- : (catalogEntry?.models ?? []);
- return entries.map((entry) => ({
- value: entry.model,
- label: entry.model,
- }));
- }, [catalogEntry, isEmbedding]);
- // Default to the first model when the catalogue loads or provider changes.
+ // Azure's catalogue lists model families, not this customer's deployments,
+ // so require an explicit choice. Other providers still default to the
+ // highest-ranked model when the catalogue loads or provider changes.
const defaultedModelRef = useRef<string | undefined>(undefined);
useEffect(() => {
- if (model || models.length === 0) return;
+ if (provider === "azure" || model || models.length === 0) return;
const defaultModel = models[0].value;
// Only set once per provider so switching back doesn't re-default.
if (defaultedModelRef.current === `${provider}:${defaultModel}`) return;
@@ -256,6 +256,11 @@ export function GenericOnboarding({
</div>
<AdvancedOnboarding
icon={<Logo className="w-4 h-4" />}
+ searchPlaceholder={
+ provider === "azure"
+ ? "Search models or type Azure deployment name"
+ : undefined
+ }
languageModels={isEmbedding ? undefined : models}
embeddingModels={isEmbedding ? models : undefined}
languageModel={isEmbedding ? undefined : model}
diff --git a/frontend/components/models/catalog-models.test.ts b/frontend/components/models/catalog-models.test.ts
index dba087bd8..388785a9e 100644
--- a/frontend/components/models/catalog-models.test.ts
+++ b/frontend/components/models/catalog-models.test.ts
@@ -7,6 +7,7 @@ import {
mergeLiveCatalogOptions,
onboardingCatalogConfigured,
onboardingCredentialFields,
+ providerCatalogOptions,
providerCredentialsSatisfied,
savedSecretFieldsForProvider,
} from "./catalog-models";
@@ -158,6 +159,75 @@ describe("groupedCatalogOptions", () => {
});
});
+describe("providerCatalogOptions", () => {
+ it("uses the settings ordering for one onboarding provider", () => {
+ const azureCatalog = {
+ providers: [
+ {
+ key: "azure",
+ name: "Azure OpenAI",
+ models: [
+ { model: "gpt-4.1", capabilities: ["function_calling"] },
+ { model: "gpt-5", capabilities: ["function_calling"] },
+ { model: "gpt-4o", capabilities: ["function_calling"] },
+ ],
+ },
+ {
+ key: "openai",
+ name: "OpenAI",
+ models: [{ model: "gpt-5.1" }],
+ },
+ ],
+ };
+
+ assert.deepEqual(
+ providerCatalogOptions(azureCatalog, "azure", "language").map(
+ (option) => option.value,
+ ),
+ ["gpt-5", "gpt-4.1", "gpt-4o"],
+ );
+ });
+
+ it("puts the three legacy Azure rows after current models without removing them", () => {
+ const azureCatalog = {
+ providers: [
+ {
+ key: "azure",
+ models: [
+ { model: "mistral-large-2402", capabilities: ["function_calling"] },
+ {
+ model: "gpt-35-turbo-16k-0613",
+ capabilities: ["function_calling"],
+ },
+ { model: "gpt-35-turbo", capabilities: ["function_calling"] },
+ { model: "gpt-5.5", capabilities: ["function_calling"] },
+ { model: "gpt-4.1", capabilities: ["function_calling"] },
+ ],
+ },
+ ],
+ };
+ const names = providerCatalogOptions(azureCatalog, "azure", "language").map(
+ (option) => option.value,
+ );
+
+ assert.deepEqual(names, [
+ "gpt-5.5",
+ "gpt-4.1",
+ "mistral-large-2402",
+ "gpt-35-turbo-16k-0613",
+ "gpt-35-turbo",
+ ]);
+ assert.deepEqual(
+ groupedCatalogOptions(
+ azureCatalog,
+ { azure: true },
+ "language",
+ )[0].options.map((option) => option.value),
+ names,
+ );
+ });
+});
+
describe("onboardingCatalogConfigured", () => {
it("does not filter the full onboarding catalogue", () => {
assert.equal(onboardingCatalogConfigured(true, true), undefined);
diff --git a/frontend/components/models/catalog-models.ts b/frontend/components/models/catalog-models.ts
index 3bfdef715..7cb0498f4 100644
--- a/frontend/components/models/catalog-models.ts
+++ b/frontend/components/models/catalog-models.ts
@@ -78,6 +78,15 @@ function toOption(entry: CatalogModel, provider: string): CatalogSelectOption {
};
}
+function sortedProviderOptions(
+ provider: CatalogProvider,
+ kind: CatalogModelKind,
+): CatalogSelectOption[] {
+ return modelsForKind(provider, kind)
+ .map((entry) => toOption(entry, provider.key))
+ .sort((left, right) => compareCatalogOptions(left, right, kind));
+}
+
/** Models onboarding actually validates (chat + tools / a real embedding). */
const PREFERRED_LANGUAGE_MODELS = [
"gpt-4o-mini",
@@ -92,6 +101,15 @@ const PREFERRED_EMBEDDING_MODELS = [
"text-embedding-3-large",
];
+// These older Azure entries sort ahead of current GPT models because the
+// generic version parser reads "2402" and "35" as larger model versions.
+// Keep them searchable, but do not feature them in the collapsed preview.
+const LOWER_PRIORITY_AZURE_MODELS = new Set([
+ "mistral-large-2402",
+ "gpt-35-turbo-16k-0613",
+ "gpt-35-turbo",
+]);
+
/** A YYYY-MM-DD or YYYYMMDD snapshot stamp inside a model id. */
const SNAPSHOT_DATE = /(20\d{2})-?(\d{2})-?(\d{2})(?!\d)/;
@@ -139,6 +157,9 @@ function optionRank(
: PREFERRED_LANGUAGE_MODELS;
const preferredIndex = preferred.indexOf(name);
return [
+ option.provider === "azure" && LOWER_PRIORITY_AZURE_MODELS.has(name)
+ ? 1
+ : 0,
// The catalogue no longer carries LiteLLM's `ft:` pricing templates, but a
// live `/models/openai` fetch returns the operator's real fine-tunes. Those
// are callable and stay listed — just never above a base model.
@@ -238,9 +259,7 @@ export function groupedCatalogOptions(
if (configured && !configured[key]) {
continue;
}
- const modelOptions = modelsForKind(provider, kind)
- .map((entry) => toOption(entry, key))
- .sort((left, right) => compareCatalogOptions(left, right, kind));
+ const modelOptions = sortedProviderOptions(provider, kind);
if (modelOptions.length === 0 && !options?.includeEmpty) {
continue;
}
@@ -253,6 +272,16 @@ export function groupedCatalogOptions(
return groups;
}
+/** One provider's models, ordered by the same ranking used in Settings. */
+export function providerCatalogOptions(
+ catalog: ModelCatalogResponse | undefined,
+ provider: string,
+ kind: CatalogModelKind,
+): CatalogSelectOption[] {
+ const entry = catalog?.providers?.find((item) => item.key === provider);
+ return entry ? sortedProviderOptions(entry, kind) : [];
+}
+
/**
* Onboarding lists every OpenRAG-credentialed provider, even before the
* operator has saved a key — they are here to configure one.
diff --git a/frontend/components/models/model-selector.test.tsx b/frontend/components/models/model-selector.test.tsx
index 2a7724ee9..4ae0c6a71 100644
--- a/frontend/components/models/model-selector.test.tsx
+++ b/frontend/components/models/model-selector.test.tsx
@@ -50,3 +50,48 @@ describe("ModelSelector custom entries", () => {
);
});
});
+
+describe("ModelSelector flat previews", () => {
+ const azureModels = [
+ ...Array.from({ length: 44 }, (_, index) => ({
+ value: `gpt-5-${String(index).padStart(2, "0")}`,
+ label: `gpt-5-${String(index).padStart(2, "0")}`,
+ })),
+ { value: "gpt-4.1", label: "gpt-4.1" },
+ ];
+
+ it("offers an explicit Show all action for onboarding models beyond the preview", async () => {
+ const user = userEvent.setup();
+ renderWithProviders(
+ <ModelSelector
+ options={azureModels}
+ previewLimit={5}
+ value=""
+ onValueChange={vi.fn()}
+ defaultOpen
+ />,
+ );
+
+ expect(screen.queryByRole("option", { name: "gpt-4.1" })).toBeNull();
+ await user.click(
+ screen.getByRole("option", { name: "Show all 45 models" }),
+ );
+ expect(screen.getByRole("option", { name: "gpt-4.1" })).toBeInTheDocument();
+ });
+
+ it("searches models beyond the preview without expanding first", async () => {
+ const user = userEvent.setup();
+ renderWithProviders(
+ <ModelSelector
+ options={azureModels}
+ previewLimit={5}
+ value=""
+ onValueChange={vi.fn()}
+ defaultOpen
+ />,
+ );
+
+ await user.type(screen.getByTestId("model-search-input"), "gpt-4.1");
+ expect(screen.getByRole("option", { name: "gpt-4.1" })).toBeInTheDocument();
+ });
+});
diff --git a/frontend/components/models/model-selector.tsx b/frontend/components/models/model-selector.tsx
index f24152065..4323ca5cc 100644
--- a/frontend/components/models/model-selector.tsx
+++ b/frontend/components/models/model-selector.tsx
@@ -90,6 +90,7 @@ export function ModelSelector({
icon,
placeholder = "Select model...",
searchPlaceholder,
+ previewLimit,
noOptionsPlaceholder = "No models available",
custom = false,
hasError = false,
@@ -117,6 +118,7 @@ export function ModelSelector({
(custom ? "Search or type a model name" : "Search model...");
const [searchValue, setSearchValue] = useState("");
+ const [showAllFlatOptions, setShowAllFlatOptions] = useState(false);
// The option filter runs on the trimmed, lowercased search text, so the
// custom entry has to use the trimmed text too — otherwise typing trailing
// whitespace both defeats the duplicate check and stores a model id with
@@ -223,11 +225,29 @@ export function ModelSelector({
const visibleOptions = useMemo(() => {
if (groupedOptions) return [];
const source = options ?? [];
- if (!deferredSearch) return source.slice(0, 40);
- return source
- .filter((option) => option.label.toLowerCase().includes(deferredSearch))
- .slice(0, 100);
- }, [deferredSearch, groupedOptions, options]);
+ if (!deferredSearch) {
+ if (previewLimit !== undefined) {
+ return showAllFlatOptions ? source : source.slice(0, previewLimit);
+ }
+ return source.slice(0, 40);
+ }
+ const matches = source.filter((option) =>
+ option.label.toLowerCase().includes(deferredSearch),
+ );
+ return previewLimit !== undefined ? matches : matches.slice(0, 100);
+ }, [
+ deferredSearch,
+ groupedOptions,
+ options,
+ previewLimit,
+ showAllFlatOptions,
+ ]);
+ const showFlatPreviewHint =
+ previewLimit !== undefined &&
+ !groupedOptions &&
+ !deferredSearch &&
+ !showAllFlatOptions &&
+ (options?.length ?? 0) > previewLimit;
// `shouldFilter={false}` means cmdk's own item count no longer reflects the
// manual filtering above, so `CommandEmpty` cannot be trusted to appear.
@@ -296,6 +316,12 @@ export function ModelSelector({
value={searchValue}
onValueChange={setSearchValue}
/>
+ {showFlatPreviewHint && (
+ <p className="px-3 py-2 text-xs text-muted-foreground">
+ Showing {previewLimit} of {options?.length} models. Search all
+ models or choose Show all.
+ </p>
+ )}
<CommandList
id={listboxId}
className="max-h-[300px] overflow-y-auto"
@@ -561,6 +587,17 @@ export function ModelSelector({
</div>
</CommandItem>
))}
+ {showFlatPreviewHint && (
+ <CommandItem
+ value="__show-all-models"
+ aria-label={`Show all ${options?.length} models`}
+ data-testid="model-show-all-options"
+ className="text-xs text-muted-foreground"
+ onSelect={() => setShowAllFlatOptions(true)}
+ >
+ Show all {options?.length} models
+ </CommandItem>
+ )}
{showFlatCustomEntry && (
<CommandItem
value={customValue}
diff --git a/frontend/components/models/types.ts b/frontend/components/models/types.ts
index 80dab834c..2421ba0a6 100644
--- a/frontend/components/models/types.ts
+++ b/frontend/components/models/types.ts
@@ -27,6 +27,8 @@ export interface ModelSelectorProps extends ButtonProps {
icon?: React.ReactNode;
placeholder?: string;
searchPlaceholder?: string;
+ /** Initial flat-list size; when set, a Show all action exposes the rest. */
+ previewLimit?: number;
noOptionsPlaceholder?: string;
custom?: boolean;
onValueChange: (value: string, provider?: string) => void;