New provider SiliconFlow

packages/inference/README.md

@@ -60,6 +60,7 @@ Currently, we support the following providers:
 - [Replicate](https://replicate.com)
 - [Sambanova](https://sambanova.ai)
 - [Scaleway](https://www.scaleway.com/en/generative-apis/)
+- [SiliconFlow](https://siliconflow.com)
 - [Clarifai](http://clarifai.com)
 - [Together](https://together.xyz)
 - [Baseten](https://baseten.co)
@@ -99,6 +100,7 @@ Only a subset of models are supported when requesting third-party providers. You
 - [Replicate supported models](https://huggingface.co/api/partners/replicate/models)
 - [Sambanova supported models](https://huggingface.co/api/partners/sambanova/models)
 - [Scaleway supported models](https://huggingface.co/api/partners/scaleway/models)
+- [SiliconFlow supported models](https://huggingface.co/api/partners/siliconflow/models)
 - [Together supported models](https://huggingface.co/api/partners/together/models)
 - [Baseten supported models](https://huggingface.co/api/partners/baseten/models)
 - [Clarifai supported models](https://huggingface.co/api/partners/clarifai/models)

packages/inference/src/lib/getProviderHelper.ts

@@ -15,6 +15,7 @@ import * as Nscale from "../providers/nscale.js";
 import * as OpenAI from "../providers/openai.js";
 import * as OvhCloud from "../providers/ovhcloud.js";
 import * as PublicAI from "../providers/publicai.js";
+import * as SiliconFlow from "../providers/siliconflow.js";
 import type {
 	AudioClassificationTaskHelper,
 	AudioToAudioTaskHelper,
@@ -169,6 +170,9 @@ export const PROVIDERS: Record<InferenceProvider, Partial<Record<InferenceTask,
 		"text-generation": new Scaleway.ScalewayTextGenerationTask(),
 		"feature-extraction": new Scaleway.ScalewayFeatureExtractionTask(),
 	},
+	siliconflow: {
+		conversational: new SiliconFlow.SiliconFlowConversationalTask(),
+	},
 	together: {
 		"text-to-image": new Together.TogetherTextToImageTask(),
 		conversational: new Together.TogetherConversationalTask(),

packages/inference/src/providers/consts.ts

@@ -38,6 +38,7 @@ export const HARDCODED_MODEL_INFERENCE_MAPPING: Record<
 	replicate: {},
 	sambanova: {},
 	scaleway: {},
+	siliconflow: {},
 	together: {},
 	wavespeed: {},
 	"zai-org": {},

packages/inference/src/providers/siliconflow.ts

@@ -0,0 +1,16 @@
+/**
+ * SiliconFlow provider implementation
+ *
+ * API Documentation: https://docs.siliconflow.com
+ *
+ * SiliconFlow follows the OpenAI API standard for LLMs.
+ */
+import { BaseConversationalTask } from "./providerHelper.js";
+
+const SILICONFLOW_API_BASE_URL = "https://api.siliconflow.com";
+
+export class SiliconFlowConversationalTask extends BaseConversationalTask {
+	constructor() {
+		super("siliconflow", SILICONFLOW_API_BASE_URL);
+	}
+}

packages/inference/src/types.ts

@@ -65,6 +65,7 @@ export const INFERENCE_PROVIDERS = [
 	"replicate",
 	"sambanova",
 	"scaleway",
+	"siliconflow",
 	"together",
 	"wavespeed",
 	"zai-org",
@@ -102,6 +103,7 @@ export const PROVIDERS_HUB_ORGS: Record<InferenceProvider, string> = {
 	replicate: "replicate",
 	sambanova: "sambanovasystems",
 	scaleway: "scaleway",
+	siliconflow: "siliconflow",
 	together: "togethercomputer",
 	wavespeed: "wavespeed",
 	"zai-org": "zai-org",

packages/inference/test/InferenceClient.spec.ts

@@ -2572,4 +2572,99 @@ describe.skip("InferenceClient", () => {
 		},
 		TIMEOUT
 	);
+	describe.concurrent(
+		"SiliconFlow",
+		() => {
+			const client = new InferenceClient(env.HF_SILICONFLOW_KEY ?? "dummy");
+
+			HARDCODED_MODEL_INFERENCE_MAPPING["siliconflow"] = {
+				"deepseek-ai/DeepSeek-R1": {
+					provider: "siliconflow",
+					hfModelId: "deepseek-ai/DeepSeek-R1",
+					providerId: "deepseek-ai/DeepSeek-R1",
+					status: "live",
+					task: "conversational",
+				},
+				"deepseek-ai/DeepSeek-V3": {
+					provider: "siliconflow",
+					hfModelId: "deepseek-ai/DeepSeek-V3",
+					providerId: "deepseek-ai/DeepSeek-V3",
+					status: "live",
+					task: "conversational",
+				},
+			};
+
+			it("chatCompletion - DeepSeek-R1", async () => {
+				const res = await client.chatCompletion({
+					model: "deepseek-ai/DeepSeek-R1",
+					provider: "siliconflow",
+					messages: [{ role: "user", content: "What is the capital of France?" }],
+					max_tokens: 20,
+				});
+				if (res.choices && res.choices.length > 0) {
+					const completion = res.choices[0].message?.content;
+					expect(completion).toBeDefined();
+					expect(typeof completion).toBe("string");
+					expect(completion).toMatch(/Paris/i);
+				}
+			});
+
+			it("chatCompletion - DeepSeek-V3", async () => {
+				const res = await client.chatCompletion({
+					model: "deepseek-ai/DeepSeek-V3",
+					provider: "siliconflow",
+					messages: [{ role: "user", content: "The weather today is" }],
+					max_tokens: 10,
+				});
+				expect(res.choices).toBeDefined();
+				expect(res.choices?.length).toBeGreaterThan(0);
+				expect(res.choices?.[0].message?.content).toBeDefined();
+				expect(typeof res.choices?.[0].message?.content).toBe("string");
+				expect(res.choices?.[0].message?.content?.length).toBeGreaterThan(0);
+			});
+
+			it("chatCompletion stream - DeepSeek-R1", async () => {
+				const stream = client.chatCompletionStream({
+					model: "deepseek-ai/DeepSeek-R1",
+					provider: "siliconflow",
+					messages: [{ role: "user", content: "Say 'this is a test'" }],
+					stream: true,
+				}) as AsyncGenerator<ChatCompletionStreamOutput>;
+
+				let fullResponse = "";
+				for await (const chunk of stream) {
+					if (chunk.choices && chunk.choices.length > 0) {
+						const content = chunk.choices[0].delta?.content;
+						if (content) {
+							fullResponse += content;
+						}
+					}
+				}
+				expect(fullResponse).toBeTruthy();
+				expect(fullResponse.length).toBeGreaterThan(0);
+			});
+
+			it("chatCompletion stream - DeepSeek-V3", async () => {
+				const stream = client.chatCompletionStream({
+					model: "deepseek-ai/DeepSeek-V3",
+					provider: "siliconflow",
+					messages: [{ role: "user", content: "Say 'this is a test'" }],
+					stream: true,
+				}) as AsyncGenerator<ChatCompletionStreamOutput>;
+
+				let fullResponse = "";
+				for await (const chunk of stream) {
+					if (chunk.choices && chunk.choices.length > 0) {
+						const content = chunk.choices[0].delta?.content;
+						if (content) {
+							fullResponse += content;
+						}
+					}
+				}
+				expect(fullResponse).toBeTruthy();
+				expect(fullResponse.length).toBeGreaterThan(0);
+			});
+		},
+		TIMEOUT
+	);
 });