feat: Add DeepInfra as an inference provider

packages/inference/README.md

@@ -66,6 +66,7 @@ Currently, we support the following providers:
 - [Blackforestlabs](https://blackforestlabs.ai)
 - [Cohere](https://cohere.com)
 - [Cerebras](https://cerebras.ai/)
+- [DeepInfra](https://deepinfra.com)
 - [Groq](https://groq.com)
 - [Wavespeed.ai](https://wavespeed.ai/)
 - [Z.ai](https://z.ai/)
@@ -104,6 +105,7 @@ Only a subset of models are supported when requesting third-party providers. You
 - [Clarifai supported models](https://huggingface.co/api/partners/clarifai/models)
 - [Cohere supported models](https://huggingface.co/api/partners/cohere/models)
 - [Cerebras supported models](https://huggingface.co/api/partners/cerebras/models)
+- [DeepInfra supported models](https://huggingface.co/api/partners/deepinfra/models)
 - [Groq supported models](https://console.groq.com/docs/models)
 - [Novita AI supported models](https://huggingface.co/api/partners/novita/models)
 - [Wavespeed.ai supported models](https://huggingface.co/api/partners/wavespeed/models)

packages/inference/src/lib/getProviderHelper.ts

@@ -3,6 +3,7 @@ import * as Clarifai from "../providers/clarifai.js";
 import * as BlackForestLabs from "../providers/black-forest-labs.js";
 import * as Cerebras from "../providers/cerebras.js";
 import * as Cohere from "../providers/cohere.js";
+import * as DeepInfra from "../providers/deepinfra.js";
 import * as FalAI from "../providers/fal-ai.js";
 import * as FeatherlessAI from "../providers/featherless-ai.js";
 import * as Fireworks from "../providers/fireworks-ai.js";
@@ -73,6 +74,10 @@ export const PROVIDERS: Record<InferenceProvider, Partial<Record<InferenceTask,
 	cohere: {
 		conversational: new Cohere.CohereConversationalTask(),
 	},
+	deepinfra: {
+		conversational: new DeepInfra.DeepInfraConversationalTask(),
+		"text-generation": new DeepInfra.DeepInfraTextGenerationTask(),
+	},
 	"fal-ai": {
 		"text-to-image": new FalAI.FalAITextToImageTask(),
 		"text-to-speech": new FalAI.FalAITextToSpeechTask(),

packages/inference/src/providers/consts.ts

@@ -23,6 +23,7 @@ export const HARDCODED_MODEL_INFERENCE_MAPPING: Record<
 	cerebras: {},
 	clarifai: {},
 	cohere: {},
+	deepinfra: {},
 	"fal-ai": {},
 	"featherless-ai": {},
 	"fireworks-ai": {},

packages/inference/src/providers/deepinfra.ts

@@ -0,0 +1,78 @@
+import type { TextGenerationOutput } from "@huggingface/tasks";
+import { InferenceClientProviderOutputError } from "../errors.js";
+import type { BodyParams } from "../types.js";
+import { omit } from "../utils/omit.js";
+import { BaseConversationalTask, BaseTextGenerationTask } from "./providerHelper.js";
+
+/**
+ * DeepInfra exposes OpenAI-compatible endpoints under the /v1/openai namespace.
+ */
+const DEEPINFRA_API_BASE_URL = "https://api.deepinfra.com";
+
+interface DeepInfraCompletionChoice {
+	text?: string;
+	message?: { content?: string };
+}
+
+interface DeepInfraCompletionResponse {
+	choices: DeepInfraCompletionChoice[];
+	model: string;
+}
+
+export class DeepInfraConversationalTask extends BaseConversationalTask {
+	constructor() {
+		super("deepinfra", DEEPINFRA_API_BASE_URL);
+	}
+
+	override makeRoute(): string {
+		return "v1/openai/chat/completions";
+	}
+}
+
+export class DeepInfraTextGenerationTask extends BaseTextGenerationTask {
+	constructor() {
+		super("deepinfra", DEEPINFRA_API_BASE_URL);
+	}
+
+	override makeRoute(): string {
+		return "v1/openai/completions";
+	}
+
+	override preparePayload(params: BodyParams): Record<string, unknown> {
+		const parameters = params.args.parameters as Record<string, unknown> | undefined;
+		const res = {
+			model: params.model,
+			prompt: params.args.inputs,
+			...omit(params.args, ["inputs", "parameters"]),
+			...(parameters
+				? {
+						max_tokens: parameters.max_new_tokens,
+						stop: parameters.stop_strings,
+						...omit(parameters, ["max_new_tokens", "stop_strings"]),
+				  }
+				: undefined),
+		};
+		return res;
+	}
+
+	override async getResponse(response: DeepInfraCompletionResponse): Promise<TextGenerationOutput> {
+		if (
+			typeof response === "object" &&
+			response !== null &&
+			Array.isArray(response.choices) &&
+			response.choices.length > 0
+		) {
+			const choice = response.choices[0];
+			const completion =
+				choice.text ??
+				(typeof choice.message?.content === "string" ? choice.message.content : undefined);
+			if (typeof completion === "string") {
+				return { generated_text: completion };
+			}
+		}
+
+		throw new InferenceClientProviderOutputError(
+			"Received malformed response from DeepInfra text-generation API: expected OpenAI completion payload"
+		);
+	}
+}

packages/inference/src/types.ts

@@ -50,6 +50,7 @@ export const INFERENCE_PROVIDERS = [
 	"cerebras",
 	"clarifai",
 	"cohere",
+	"deepinfra",
 	"fal-ai",
 	"featherless-ai",
 	"fireworks-ai",

packages/inference/test/InferenceClient.spec.ts

@@ -1967,6 +1967,82 @@ describe.skip("InferenceClient", () => {
 		},
 		TIMEOUT
 	);
+	describe.concurrent(
+		"DeepInfra",
+		() => {
+			const client = new InferenceClient(
+				env.DEEPINFRA_API_KEY ?? env.HF_DEEPINFRA_KEY ?? "dummy"
+			);
+
+			const HF_MODEL = "google/gemma-3-4b-it";
+			const PROVIDER_ID = "google/gemma-3-4b-it";
+
+			const setMapping = (task: "conversational" | "text-generation") => {
+				HARDCODED_MODEL_INFERENCE_MAPPING["deepinfra"] = {
+					[HF_MODEL]: {
+						provider: "deepinfra",
+						hfModelId: HF_MODEL,
+						providerId: PROVIDER_ID,
+						status: "live",
+						task,
+					},
+				};
+			};
+
+			it("chatCompletion", async () => {
+				setMapping("conversational");
+				const res = await client.chatCompletion({
+					model: HF_MODEL,
+					provider: "deepinfra",
+					messages: [{ role: "user", content: "Name one use case for open-source AI." }],
+					max_tokens: 64,
+				});
+				if (res.choices && res.choices.length > 0) {
+					const completion = res.choices[0].message?.content;
+					expect(typeof completion).toBe("string");
+					expect(completion?.length ?? 0).toBeGreaterThan(0);
+				}
+			});
+
+			it("chatCompletion stream", async () => {
+				setMapping("conversational");
+				const stream = client.chatCompletionStream({
+					model: HF_MODEL,
+					provider: "deepinfra",
+					messages: [
+						{
+							role: "user",
+							content: "Respond with a two-word description of Hugging Face.",
+						},
+					],
+					max_tokens: 32,
+				}) as AsyncGenerator<ChatCompletionStreamOutput>;
+				let streamed = "";
+				for await (const chunk of stream) {
+					if (chunk.choices && chunk.choices.length > 0) {
+						const delta = chunk.choices[0].delta?.content;
+						if (delta) {
+							streamed += delta;
+						}
+					}
+				}
+				expect(streamed.trim().length).toBeGreaterThan(0);
+			});
+
+			it("textGeneration", async () => {
+				setMapping("text-generation");
+				const generation = await client.textGeneration({
+					model: HF_MODEL,
+					provider: "deepinfra",
+					inputs: "Describe Hugging Face in one short sentence.",
+					parameters: { max_new_tokens: 64 },
+				});
+				expect(typeof generation.generated_text).toBe("string");
+				expect(generation.generated_text?.length ?? 0).toBeGreaterThan(0);
+			});
+		},
+		TIMEOUT
+	);
 	describe.concurrent(
 		"Cerebras",
 		() => {

packages/tasks/src/inference-providers.ts

@@ -3,6 +3,7 @@
 const INFERENCE_PROVIDERS = [
 	"cerebras",
 	"cohere",
+	"deepinfra",
 	"fal-ai",
 	"fireworks-ai",
 	"hf-inference",