@huggingface/inference 3.7.1 → 3.8.1

package/dist/index.cjs

@@ -98,15 +98,6 @@ __export(tasks_exports, {
   zeroShotImageClassification: () => zeroShotImageClassification
 });
 
-// package.json
-var name = "@huggingface/inference";
-var version = "3.7.1";
-
-// src/config.ts
-var HF_HUB_URL = "https://huggingface.co";
-var HF_ROUTER_URL = "https://router.huggingface.co";
-var HF_HEADER_X_BILL_TO = "X-HF-Bill-To";
-
 // src/lib/InferenceOutputError.ts
 var InferenceOutputError = class extends TypeError {
   constructor(message) {
@@ -148,6 +139,11 @@ function omit(o, props) {
   return pick(o, letsKeep);
 }
 
+// src/config.ts
+var HF_HUB_URL = "https://huggingface.co";
+var HF_ROUTER_URL = "https://router.huggingface.co";
+var HF_HEADER_X_BILL_TO = "X-HF-Bill-To";
+
 // src/utils/toArray.ts
 function toArray(obj) {
   if (Array.isArray(obj)) {
@@ -337,14 +333,37 @@ var FalAITask = class extends TaskProviderHelper {
     return headers;
   }
 };
+function buildLoraPath(modelId, adapterWeightsPath) {
+  return `${HF_HUB_URL}/${modelId}/resolve/main/${adapterWeightsPath}`;
+}
 var FalAITextToImageTask = class extends FalAITask {
   preparePayload(params) {
-    return {
+    const payload = {
       ...omit(params.args, ["inputs", "parameters"]),
       ...params.args.parameters,
       sync_mode: true,
-      prompt: params.args.inputs
+      prompt: params.args.inputs,
+      ...params.mapping?.adapter === "lora" && params.mapping.adapterWeightsPath ? {
+        loras: [
+          {
+            path: buildLoraPath(params.mapping.hfModelId, params.mapping.adapterWeightsPath),
+            scale: 1
+          }
+        ]
+      } : void 0
     };
+    if (params.mapping?.adapter === "lora" && params.mapping.adapterWeightsPath) {
+      payload.loras = [
+        {
+          path: buildLoraPath(params.mapping.hfModelId, params.mapping.adapterWeightsPath),
+          scale: 1
+        }
+      ];
+      if (params.mapping.providerId === "fal-ai/lora") {
+        payload.model_name = "stabilityai/stable-diffusion-xl-base-1.0";
+      }
+    }
+    return payload;
   }
   async getResponse(response, outputType) {
     if (typeof response === "object" && "images" in response && Array.isArray(response.images) && response.images.length > 0 && "url" in response.images[0] && typeof response.images[0].url === "string") {
@@ -474,6 +493,7 @@ var FireworksConversationalTask = class extends BaseConversationalTask {
 };
 
 // src/providers/hf-inference.ts
+var EQUIVALENT_SENTENCE_TRANSFORMERS_TASKS = ["feature-extraction", "sentence-similarity"];
 var HFInferenceTask = class extends TaskProviderHelper {
   constructor() {
     super("hf-inference", `${HF_ROUTER_URL}/hf-inference`);
@@ -1181,8 +1201,12 @@ function getProviderHelper(provider, task) {
   return providerTasks[task];
 }
 
+// package.json
+var name = "@huggingface/inference";
+var version = "3.8.1";
+
 // src/providers/consts.ts
-var HARDCODED_MODEL_ID_MAPPING = {
+var HARDCODED_MODEL_INFERENCE_MAPPING = {
   /**
    * "HF model ID" => "Model ID on Inference Provider's side"
    *
@@ -1204,53 +1228,67 @@ var HARDCODED_MODEL_ID_MAPPING = {
   together: {}
 };
 
-// src/lib/getProviderModelId.ts
+// src/lib/getInferenceProviderMapping.ts
 var inferenceProviderMappingCache = /* @__PURE__ */ new Map();
-async function getProviderModelId(params, args, options = {}) {
-  if (params.provider === "hf-inference") {
-    return params.model;
-  }
-  if (!options.task) {
-    throw new Error("task must be specified when using a third-party provider");
-  }
-  const task = options.task === "text-generation" && options.chatCompletion ? "conversational" : options.task;
-  if (HARDCODED_MODEL_ID_MAPPING[params.provider]?.[params.model]) {
-    return HARDCODED_MODEL_ID_MAPPING[params.provider][params.model];
+async function getInferenceProviderMapping(params, options) {
+  if (HARDCODED_MODEL_INFERENCE_MAPPING[params.provider][params.modelId]) {
+    return HARDCODED_MODEL_INFERENCE_MAPPING[params.provider][params.modelId];
   }
   let inferenceProviderMapping;
-  if (inferenceProviderMappingCache.has(params.model)) {
-    inferenceProviderMapping = inferenceProviderMappingCache.get(params.model);
+  if (inferenceProviderMappingCache.has(params.modelId)) {
+    inferenceProviderMapping = inferenceProviderMappingCache.get(params.modelId);
   } else {
-    inferenceProviderMapping = await (options?.fetch ?? fetch)(
-      `${HF_HUB_URL}/api/models/${params.model}?expand[]=inferenceProviderMapping`,
+    const resp = await (options?.fetch ?? fetch)(
+      `${HF_HUB_URL}/api/models/${params.modelId}?expand[]=inferenceProviderMapping`,
       {
-        headers: args.accessToken?.startsWith("hf_") ? { Authorization: `Bearer ${args.accessToken}` } : {}
+        headers: params.accessToken?.startsWith("hf_") ? { Authorization: `Bearer ${params.accessToken}` } : {}
       }
-    ).then((resp) => resp.json()).then((json) => json.inferenceProviderMapping).catch(() => null);
+    );
+    if (resp.status === 404) {
+      throw new Error(`Model ${params.modelId} does not exist`);
+    }
+    inferenceProviderMapping = await resp.json().then((json) => json.inferenceProviderMapping).catch(() => null);
   }
   if (!inferenceProviderMapping) {
-    throw new Error(`We have not been able to find inference provider information for model ${params.model}.`);
+    throw new Error(`We have not been able to find inference provider information for model ${params.modelId}.`);
   }
   const providerMapping = inferenceProviderMapping[params.provider];
   if (providerMapping) {
-    if (providerMapping.task !== task) {
+    const equivalentTasks = params.provider === "hf-inference" && typedInclude(EQUIVALENT_SENTENCE_TRANSFORMERS_TASKS, params.task) ? EQUIVALENT_SENTENCE_TRANSFORMERS_TASKS : [params.task];
+    if (!typedInclude(equivalentTasks, providerMapping.task)) {
       throw new Error(
-        `Model ${params.model} is not supported for task ${task} and provider ${params.provider}. Supported task: ${providerMapping.task}.`
+        `Model ${params.modelId} is not supported for task ${params.task} and provider ${params.provider}. Supported task: ${providerMapping.task}.`
       );
     }
     if (providerMapping.status === "staging") {
       console.warn(
-        `Model ${params.model} is in staging mode for provider ${params.provider}. Meant for test purposes only.`
+        `Model ${params.modelId} is in staging mode for provider ${params.provider}. Meant for test purposes only.`
       );
     }
-    return providerMapping.providerId;
+    if (providerMapping.adapter === "lora") {
+      const treeResp = await (options?.fetch ?? fetch)(`${HF_HUB_URL}/api/models/${params.modelId}/tree/main`);
+      if (!treeResp.ok) {
+        throw new Error(`Unable to fetch the model tree for ${params.modelId}.`);
+      }
+      const tree = await treeResp.json();
+      const adapterWeightsPath = tree.find(({ type, path }) => type === "file" && path.endsWith(".safetensors"))?.path;
+      if (!adapterWeightsPath) {
+        throw new Error(`No .safetensors file found in the model tree for ${params.modelId}.`);
+      }
+      return {
+        ...providerMapping,
+        hfModelId: params.modelId,
+        adapterWeightsPath
+      };
+    }
+    return { ...providerMapping, hfModelId: params.modelId };
   }
-  throw new Error(`Model ${params.model} is not supported provider ${params.provider}.`);
+  return null;
 }
 
 // src/lib/makeRequestOptions.ts
 var tasks = null;
-async function makeRequestOptions(args, options) {
+async function makeRequestOptions(args, providerHelper, options) {
   const { provider: maybeProvider, model: maybeModel } = args;
   const provider = maybeProvider ?? "hf-inference";
   const { task } = options ?? {};
@@ -1260,28 +1298,55 @@ async function makeRequestOptions(args, options) {
   if (maybeModel && isUrl(maybeModel)) {
     throw new Error(`Model URLs are no longer supported. Use endpointUrl instead.`);
   }
+  if (args.endpointUrl) {
+    return makeRequestOptionsFromResolvedModel(
+      maybeModel ?? args.endpointUrl,
+      providerHelper,
+      args,
+      void 0,
+      options
+    );
+  }
   if (!maybeModel && !task) {
     throw new Error("No model provided, and no task has been specified.");
   }
   const hfModel = maybeModel ?? await loadDefaultModel(task);
-  const providerHelper = getProviderHelper(provider, task);
   if (providerHelper.clientSideRoutingOnly && !maybeModel) {
     throw new Error(`Provider ${provider} requires a model ID to be passed directly.`);
   }
-  const resolvedModel = providerHelper.clientSideRoutingOnly ? (
+  const inferenceProviderMapping = providerHelper.clientSideRoutingOnly ? {
     // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
-    removeProviderPrefix(maybeModel, provider)
-  ) : await getProviderModelId({ model: hfModel, provider }, args, {
-    task,
-    fetch: options?.fetch
-  });
-  return makeRequestOptionsFromResolvedModel(resolvedModel, args, options);
+    providerId: removeProviderPrefix(maybeModel, provider),
+    // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
+    hfModelId: maybeModel,
+    status: "live",
+    // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
+    task
+  } : await getInferenceProviderMapping(
+    {
+      modelId: hfModel,
+      // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
+      task,
+      provider,
+      accessToken: args.accessToken
+    },
+    { fetch: options?.fetch }
+  );
+  if (!inferenceProviderMapping) {
+    throw new Error(`We have not been able to find inference provider information for model ${hfModel}.`);
+  }
+  return makeRequestOptionsFromResolvedModel(
+    inferenceProviderMapping.providerId,
+    providerHelper,
+    args,
+    inferenceProviderMapping,
+    options
+  );
 }
-function makeRequestOptionsFromResolvedModel(resolvedModel, args, options) {
+function makeRequestOptionsFromResolvedModel(resolvedModel, providerHelper, args, mapping, options) {
   const { accessToken, endpointUrl, provider: maybeProvider, model, ...remainingArgs } = args;
   const provider = maybeProvider ?? "hf-inference";
   const { includeCredentials, task, signal, billTo } = options ?? {};
-  const providerHelper = getProviderHelper(provider, task);
   const authMethod = (() => {
     if (providerHelper.clientSideRoutingOnly) {
       if (accessToken && accessToken.startsWith("hf_")) {
@@ -1319,7 +1384,8 @@ function makeRequestOptionsFromResolvedModel(resolvedModel, args, options) {
   const body = providerHelper.makeBody({
     args: remainingArgs,
     model: resolvedModel,
-    task
+    task,
+    mapping
   });
   let credentials;
   if (typeof includeCredentials === "string") {
@@ -1460,12 +1526,12 @@ function newMessage() {
 }
 
 // src/utils/request.ts
-async function innerRequest(args, options) {
-  const { url, info } = await makeRequestOptions(args, options);
+async function innerRequest(args, providerHelper, options) {
+  const { url, info } = await makeRequestOptions(args, providerHelper, options);
   const response = await (options?.fetch ?? fetch)(url, info);
   const requestContext = { url, info };
   if (options?.retry_on_error !== false && response.status === 503) {
-    return innerRequest(args, options);
+    return innerRequest(args, providerHelper, options);
   }
   if (!response.ok) {
     const contentType = response.headers.get("Content-Type");
@@ -1492,11 +1558,11 @@ async function innerRequest(args, options) {
   const blob = await response.blob();
   return { data: blob, requestContext };
 }
-async function* innerStreamingRequest(args, options) {
-  const { url, info } = await makeRequestOptions({ ...args, stream: true }, options);
+async function* innerStreamingRequest(args, providerHelper, options) {
+  const { url, info } = await makeRequestOptions({ ...args, stream: true }, providerHelper, options);
   const response = await (options?.fetch ?? fetch)(url, info);
   if (options?.retry_on_error !== false && response.status === 503) {
-    return yield* innerStreamingRequest(args, options);
+    return yield* innerStreamingRequest(args, providerHelper, options);
   }
   if (!response.ok) {
     if (response.headers.get("Content-Type")?.startsWith("application/json")) {
@@ -1570,7 +1636,8 @@ async function request(args, options) {
   console.warn(
     "The request method is deprecated and will be removed in a future version of huggingface.js. Use specific task functions instead."
   );
-  const result = await innerRequest(args, options);
+  const providerHelper = getProviderHelper(args.provider ?? "hf-inference", options?.task);
+  const result = await innerRequest(args, providerHelper, options);
   return result.data;
 }
 
@@ -1579,7 +1646,8 @@ async function* streamingRequest(args, options) {
   console.warn(
     "The streamingRequest method is deprecated and will be removed in a future version of huggingface.js. Use specific task functions instead."
   );
-  yield* innerStreamingRequest(args, options);
+  const providerHelper = getProviderHelper(args.provider ?? "hf-inference", options?.task);
+  yield* innerStreamingRequest(args, providerHelper, options);
 }
 
 // src/tasks/audio/utils.ts
@@ -1594,7 +1662,7 @@ function preparePayload(args) {
 async function audioClassification(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "audio-classification");
   const payload = preparePayload(args);
-  const { data: res } = await innerRequest(payload, {
+  const { data: res } = await innerRequest(payload, providerHelper, {
     ...options,
     task: "audio-classification"
   });
@@ -1605,7 +1673,7 @@ async function audioClassification(args, options) {
 async function audioToAudio(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "audio-to-audio");
   const payload = preparePayload(args);
-  const { data: res } = await innerRequest(payload, {
+  const { data: res } = await innerRequest(payload, providerHelper, {
     ...options,
     task: "audio-to-audio"
   });
@@ -1629,7 +1697,7 @@ function base64FromBytes(arr) {
 async function automaticSpeechRecognition(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "automatic-speech-recognition");
   const payload = await buildPayload(args);
-  const { data: res } = await innerRequest(payload, {
+  const { data: res } = await innerRequest(payload, providerHelper, {
     ...options,
     task: "automatic-speech-recognition"
   });
@@ -1669,7 +1737,7 @@ async function buildPayload(args) {
 async function textToSpeech(args, options) {
   const provider = args.provider ?? "hf-inference";
   const providerHelper = getProviderHelper(provider, "text-to-speech");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "text-to-speech"
   });
@@ -1685,7 +1753,7 @@ function preparePayload2(args) {
 async function imageClassification(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "image-classification");
   const payload = preparePayload2(args);
-  const { data: res } = await innerRequest(payload, {
+  const { data: res } = await innerRequest(payload, providerHelper, {
     ...options,
     task: "image-classification"
   });
@@ -1696,7 +1764,7 @@ async function imageClassification(args, options) {
 async function imageSegmentation(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "image-segmentation");
   const payload = preparePayload2(args);
-  const { data: res } = await innerRequest(payload, {
+  const { data: res } = await innerRequest(payload, providerHelper, {
     ...options,
     task: "image-segmentation"
   });
@@ -1721,7 +1789,7 @@ async function imageToImage(args, options) {
       )
     };
   }
-  const { data: res } = await innerRequest(reqArgs, {
+  const { data: res } = await innerRequest(reqArgs, providerHelper, {
     ...options,
     task: "image-to-image"
   });
@@ -1732,7 +1800,7 @@ async function imageToImage(args, options) {
 async function imageToText(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "image-to-text");
   const payload = preparePayload2(args);
-  const { data: res } = await innerRequest(payload, {
+  const { data: res } = await innerRequest(payload, providerHelper, {
     ...options,
     task: "image-to-text"
   });
@@ -1743,7 +1811,7 @@ async function imageToText(args, options) {
 async function objectDetection(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "object-detection");
   const payload = preparePayload2(args);
-  const { data: res } = await innerRequest(payload, {
+  const { data: res } = await innerRequest(payload, providerHelper, {
     ...options,
     task: "object-detection"
   });
@@ -1754,11 +1822,11 @@ async function objectDetection(args, options) {
 async function textToImage(args, options) {
   const provider = args.provider ?? "hf-inference";
   const providerHelper = getProviderHelper(provider, "text-to-image");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "text-to-image"
   });
-  const { url, info } = await makeRequestOptions(args, { ...options, task: "text-to-image" });
+  const { url, info } = await makeRequestOptions(args, providerHelper, { ...options, task: "text-to-image" });
   return providerHelper.getResponse(res, url, info.headers, options?.outputType);
 }
 
@@ -1766,11 +1834,15 @@ async function textToImage(args, options) {
 async function textToVideo(args, options) {
   const provider = args.provider ?? "hf-inference";
   const providerHelper = getProviderHelper(provider, "text-to-video");
-  const { data: response } = await innerRequest(args, {
-    ...options,
-    task: "text-to-video"
-  });
-  const { url, info } = await makeRequestOptions(args, { ...options, task: "text-to-video" });
+  const { data: response } = await innerRequest(
+    args,
+    providerHelper,
+    {
+      ...options,
+      task: "text-to-video"
+    }
+  );
+  const { url, info } = await makeRequestOptions(args, providerHelper, { ...options, task: "text-to-video" });
   return providerHelper.getResponse(response, url, info.headers);
 }
 
@@ -1799,7 +1871,7 @@ async function preparePayload3(args) {
 async function zeroShotImageClassification(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "zero-shot-image-classification");
   const payload = await preparePayload3(args);
-  const { data: res } = await innerRequest(payload, {
+  const { data: res } = await innerRequest(payload, providerHelper, {
     ...options,
     task: "zero-shot-image-classification"
   });
@@ -1809,7 +1881,7 @@ async function zeroShotImageClassification(args, options) {
 // src/tasks/nlp/chatCompletion.ts
 async function chatCompletion(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "conversational");
-  const { data: response } = await innerRequest(args, {
+  const { data: response } = await innerRequest(args, providerHelper, {
     ...options,
     task: "conversational"
   });
@@ -1818,7 +1890,8 @@ async function chatCompletion(args, options) {
 
 // src/tasks/nlp/chatCompletionStream.ts
 async function* chatCompletionStream(args, options) {
-  yield* innerStreamingRequest(args, {
+  const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "conversational");
+  yield* innerStreamingRequest(args, providerHelper, {
     ...options,
     task: "conversational"
   });
@@ -1827,7 +1900,7 @@ async function* chatCompletionStream(args, options) {
 // src/tasks/nlp/featureExtraction.ts
 async function featureExtraction(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "feature-extraction");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "feature-extraction"
   });
@@ -1837,7 +1910,7 @@ async function featureExtraction(args, options) {
 // src/tasks/nlp/fillMask.ts
 async function fillMask(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "fill-mask");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "fill-mask"
   });
@@ -1847,17 +1920,21 @@ async function fillMask(args, options) {
 // src/tasks/nlp/questionAnswering.ts
 async function questionAnswering(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "question-answering");
-  const { data: res } = await innerRequest(args, {
-    ...options,
-    task: "question-answering"
-  });
+  const { data: res } = await innerRequest(
+    args,
+    providerHelper,
+    {
+      ...options,
+      task: "question-answering"
+    }
+  );
   return providerHelper.getResponse(res);
 }
 
 // src/tasks/nlp/sentenceSimilarity.ts
 async function sentenceSimilarity(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "sentence-similarity");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "sentence-similarity"
   });
@@ -1867,7 +1944,7 @@ async function sentenceSimilarity(args, options) {
 // src/tasks/nlp/summarization.ts
 async function summarization(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "summarization");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "summarization"
   });
@@ -1877,17 +1954,21 @@ async function summarization(args, options) {
 // src/tasks/nlp/tableQuestionAnswering.ts
 async function tableQuestionAnswering(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "table-question-answering");
-  const { data: res } = await innerRequest(args, {
-    ...options,
-    task: "table-question-answering"
-  });
+  const { data: res } = await innerRequest(
+    args,
+    providerHelper,
+    {
+      ...options,
+      task: "table-question-answering"
+    }
+  );
   return providerHelper.getResponse(res);
 }
 
 // src/tasks/nlp/textClassification.ts
 async function textClassification(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "text-classification");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "text-classification"
   });
@@ -1896,9 +1977,8 @@ async function textClassification(args, options) {
 
 // src/tasks/nlp/textGeneration.ts
 async function textGeneration(args, options) {
-  const provider = args.provider ?? "hf-inference";
-  const providerHelper = getProviderHelper(provider, "text-generation");
-  const { data: response } = await innerRequest(args, {
+  const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "text-generation");
+  const { data: response } = await innerRequest(args, providerHelper, {
     ...options,
     task: "text-generation"
   });
@@ -1907,7 +1987,8 @@ async function textGeneration(args, options) {
 
 // src/tasks/nlp/textGenerationStream.ts
 async function* textGenerationStream(args, options) {
-  yield* innerStreamingRequest(args, {
+  const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "text-generation");
+  yield* innerStreamingRequest(args, providerHelper, {
     ...options,
     task: "text-generation"
   });
@@ -1916,17 +1997,21 @@ async function* textGenerationStream(args, options) {
 // src/tasks/nlp/tokenClassification.ts
 async function tokenClassification(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "token-classification");
-  const { data: res } = await innerRequest(args, {
-    ...options,
-    task: "token-classification"
-  });
+  const { data: res } = await innerRequest(
+    args,
+    providerHelper,
+    {
+      ...options,
+      task: "token-classification"
+    }
+  );
   return providerHelper.getResponse(res);
 }
 
 // src/tasks/nlp/translation.ts
 async function translation(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "translation");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "translation"
   });
@@ -1936,10 +2021,14 @@ async function translation(args, options) {
 // src/tasks/nlp/zeroShotClassification.ts
 async function zeroShotClassification(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "zero-shot-classification");
-  const { data: res } = await innerRequest(args, {
-    ...options,
-    task: "zero-shot-classification"
-  });
+  const { data: res } = await innerRequest(
+    args,
+    providerHelper,
+    {
+      ...options,
+      task: "zero-shot-classification"
+    }
+  );
   return providerHelper.getResponse(res);
 }
 
@@ -1956,6 +2045,7 @@ async function documentQuestionAnswering(args, options) {
   };
   const { data: res } = await innerRequest(
     reqArgs,
+    providerHelper,
     {
       ...options,
       task: "document-question-answering"
@@ -1975,7 +2065,7 @@ async function visualQuestionAnswering(args, options) {
       image: base64FromBytes(new Uint8Array(await args.inputs.image.arrayBuffer()))
     }
   };
-  const { data: res } = await innerRequest(reqArgs, {
+  const { data: res } = await innerRequest(reqArgs, providerHelper, {
     ...options,
     task: "visual-question-answering"
   });
@@ -1985,7 +2075,7 @@ async function visualQuestionAnswering(args, options) {
 // src/tasks/tabular/tabularClassification.ts
 async function tabularClassification(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "tabular-classification");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "tabular-classification"
   });
@@ -1995,7 +2085,7 @@ async function tabularClassification(args, options) {
 // src/tasks/tabular/tabularRegression.ts
 async function tabularRegression(args, options) {
   const providerHelper = getProviderHelper(args.provider ?? "hf-inference", "tabular-regression");
-  const { data: res } = await innerRequest(args, {
+  const { data: res } = await innerRequest(args, providerHelper, {
     ...options,
     task: "tabular-regression"
   });
@@ -2075,19 +2165,19 @@ var import_tasks = require("@huggingface/tasks");
 var templates = {
   "js": {
     "fetch": {
-      "basic": 'async function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "application/json",\n			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.json();\n	return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    console.log(JSON.stringify(response));\n});',
-      "basicAudio": 'async function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "audio/flac"\n			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.json();\n	return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    console.log(JSON.stringify(response));\n});',
-      "basicImage": 'async function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "image/jpeg"\n			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.json();\n	return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    console.log(JSON.stringify(response));\n});',
-      "textToAudio": '{% if model.library_name == "transformers" %}\nasync function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "application/json",\n			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.blob();\n    return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    // Returns a byte object of the Audio wavform. Use it directly!\n});\n{% else %}\nasync function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "application/json",\n			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n    const result = await response.json();\n    return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    console.log(JSON.stringify(response));\n});\n{% endif %} ',
-      "textToImage": 'async function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "application/json",\n			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.blob();\n	return result;\n}\n\n\nquery({ {{ providerInputs.asTsString }} }).then((response) => {\n    // Use image\n});',
-      "zeroShotClassification": 'async function query(data) {\n    const response = await fetch(\n		"{{ fullUrl }}",\n        {\n            headers: {\n				Authorization: "{{ authorizationHeader }}",\n                "Content-Type": "application/json",\n            },\n            method: "POST",\n            body: JSON.stringify(data),\n        }\n    );\n    const result = await response.json();\n    return result;\n}\n\nquery({\n    inputs: {{ providerInputs.asObj.inputs }},\n    parameters: { candidate_labels: ["refund", "legal", "faq"] }\n}).then((response) => {\n    console.log(JSON.stringify(response));\n});'
+      "basic": 'async function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "application/json",\n{% if billTo %}\n				"X-HF-Bill-To": "{{ billTo }}",\n{% endif %}			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.json();\n	return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    console.log(JSON.stringify(response));\n});',
+      "basicAudio": 'async function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "audio/flac",\n{% if billTo %}\n				"X-HF-Bill-To": "{{ billTo }}",\n{% endif %}			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.json();\n	return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    console.log(JSON.stringify(response));\n});',
+      "basicImage": 'async function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "image/jpeg",\n{% if billTo %}\n				"X-HF-Bill-To": "{{ billTo }}",\n{% endif %}			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.json();\n	return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    console.log(JSON.stringify(response));\n});',
+      "textToAudio": '{% if model.library_name == "transformers" %}\nasync function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "application/json",\n{% if billTo %}\n				"X-HF-Bill-To": "{{ billTo }}",\n{% endif %}			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.blob();\n    return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    // Returns a byte object of the Audio wavform. Use it directly!\n});\n{% else %}\nasync function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "application/json",\n			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n    const result = await response.json();\n    return result;\n}\n\nquery({ inputs: {{ providerInputs.asObj.inputs }} }).then((response) => {\n    console.log(JSON.stringify(response));\n});\n{% endif %} ',
+      "textToImage": 'async function query(data) {\n	const response = await fetch(\n		"{{ fullUrl }}",\n		{\n			headers: {\n				Authorization: "{{ authorizationHeader }}",\n				"Content-Type": "application/json",\n{% if billTo %}\n				"X-HF-Bill-To": "{{ billTo }}",\n{% endif %}			},\n			method: "POST",\n			body: JSON.stringify(data),\n		}\n	);\n	const result = await response.blob();\n	return result;\n}\n\n\nquery({ {{ providerInputs.asTsString }} }).then((response) => {\n    // Use image\n});',
+      "zeroShotClassification": 'async function query(data) {\n    const response = await fetch(\n		"{{ fullUrl }}",\n        {\n            headers: {\n				Authorization: "{{ authorizationHeader }}",\n                "Content-Type": "application/json",\n{% if billTo %}\n                "X-HF-Bill-To": "{{ billTo }}",\n{% endif %}         },\n            method: "POST",\n            body: JSON.stringify(data),\n        }\n    );\n    const result = await response.json();\n    return result;\n}\n\nquery({\n    inputs: {{ providerInputs.asObj.inputs }},\n    parameters: { candidate_labels: ["refund", "legal", "faq"] }\n}).then((response) => {\n    console.log(JSON.stringify(response));\n});'
     },
     "huggingface.js": {
-      "basic": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nconst output = await client.{{ methodName }}({\n	model: "{{ model.id }}",\n	inputs: {{ inputs.asObj.inputs }},\n	provider: "{{ provider }}",\n});\n\nconsole.log(output);',
-      "basicAudio": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nconst data = fs.readFileSync({{inputs.asObj.inputs}});\n\nconst output = await client.{{ methodName }}({\n	data,\n	model: "{{ model.id }}",\n	provider: "{{ provider }}",\n});\n\nconsole.log(output);',
-      "basicImage": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nconst data = fs.readFileSync({{inputs.asObj.inputs}});\n\nconst output = await client.{{ methodName }}({\n	data,\n	model: "{{ model.id }}",\n	provider: "{{ provider }}",\n});\n\nconsole.log(output);',
-      "conversational": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nconst chatCompletion = await client.chatCompletion({\n    provider: "{{ provider }}",\n    model: "{{ model.id }}",\n{{ inputs.asTsString }}\n});\n\nconsole.log(chatCompletion.choices[0].message);',
-      "conversationalStream": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nlet out = "";\n\nconst stream = await client.chatCompletionStream({\n    provider: "{{ provider }}",\n    model: "{{ model.id }}",\n{{ inputs.asTsString }}\n});\n\nfor await (const chunk of stream) {\n	if (chunk.choices && chunk.choices.length > 0) {\n		const newContent = chunk.choices[0].delta.content;\n		out += newContent;\n		console.log(newContent);\n	}  \n}',
+      "basic": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nconst output = await client.{{ methodName }}({\n	model: "{{ model.id }}",\n	inputs: {{ inputs.asObj.inputs }},\n	provider: "{{ provider }}",\n}{% if billTo %}, {\n	billTo: "{{ billTo }}",\n}{% endif %});\n\nconsole.log(output);',
+      "basicAudio": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nconst data = fs.readFileSync({{inputs.asObj.inputs}});\n\nconst output = await client.{{ methodName }}({\n	data,\n	model: "{{ model.id }}",\n	provider: "{{ provider }}",\n}{% if billTo %}, {\n	billTo: "{{ billTo }}",\n}{% endif %});\n\nconsole.log(output);',
+      "basicImage": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nconst data = fs.readFileSync({{inputs.asObj.inputs}});\n\nconst output = await client.{{ methodName }}({\n	data,\n	model: "{{ model.id }}",\n	provider: "{{ provider }}",\n}{% if billTo %}, {\n	billTo: "{{ billTo }}",\n}{% endif %});\n\nconsole.log(output);',
+      "conversational": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nconst chatCompletion = await client.chatCompletion({\n    provider: "{{ provider }}",\n    model: "{{ model.id }}",\n{{ inputs.asTsString }}\n}{% if billTo %}, {\n    billTo: "{{ billTo }}",\n}{% endif %});\n\nconsole.log(chatCompletion.choices[0].message);',
+      "conversationalStream": 'import { InferenceClient } from "@huggingface/inference";\n\nconst client = new InferenceClient("{{ accessToken }}");\n\nlet out = "";\n\nconst stream = await client.chatCompletionStream({\n    provider: "{{ provider }}",\n    model: "{{ model.id }}",\n{{ inputs.asTsString }}\n}{% if billTo %}, {\n    billTo: "{{ billTo }}",\n}{% endif %});\n\nfor await (const chunk of stream) {\n	if (chunk.choices && chunk.choices.length > 0) {\n		const newContent = chunk.choices[0].delta.content;\n		out += newContent;\n		console.log(newContent);\n	}  \n}',
       "textToImage": `import { InferenceClient } from "@huggingface/inference";
 
 const client = new InferenceClient("{{ accessToken }}");
@@ -2097,7 +2187,9 @@ const image = await client.textToImage({
     model: "{{ model.id }}",
 	inputs: {{ inputs.asObj.inputs }},
 	parameters: { num_inference_steps: 5 },
-});
+}{% if billTo %}, {
+    billTo: "{{ billTo }}",
+}{% endif %});
 /// Use the generated image (it's a Blob)`,
       "textToVideo": `import { InferenceClient } from "@huggingface/inference";
 
@@ -2107,12 +2199,14 @@ const image = await client.textToVideo({
     provider: "{{ provider }}",
     model: "{{ model.id }}",
 	inputs: {{ inputs.asObj.inputs }},
-});
+}{% if billTo %}, {
+    billTo: "{{ billTo }}",
+}{% endif %});
 // Use the generated video (it's a Blob)`
     },
     "openai": {
-      "conversational": 'import { OpenAI } from "openai";\n\nconst client = new OpenAI({\n	baseURL: "{{ baseUrl }}",\n	apiKey: "{{ accessToken }}",\n});\n\nconst chatCompletion = await client.chat.completions.create({\n	model: "{{ providerModelId }}",\n{{ inputs.asTsString }}\n});\n\nconsole.log(chatCompletion.choices[0].message);',
-      "conversationalStream": 'import { OpenAI } from "openai";\n\nconst client = new OpenAI({\n	baseURL: "{{ baseUrl }}",\n	apiKey: "{{ accessToken }}",\n});\n\nconst stream = await client.chat.completions.create({\n    model: "{{ providerModelId }}",\n{{ inputs.asTsString }}\n    stream: true,\n});\n\nfor await (const chunk of stream) {\n    process.stdout.write(chunk.choices[0]?.delta?.content || "");\n}'
+      "conversational": 'import { OpenAI } from "openai";\n\nconst client = new OpenAI({\n	baseURL: "{{ baseUrl }}",\n	apiKey: "{{ accessToken }}",\n{% if billTo %}\n	defaultHeaders: {\n		"X-HF-Bill-To": "{{ billTo }}" \n	}\n{% endif %}\n});\n\nconst chatCompletion = await client.chat.completions.create({\n	model: "{{ providerModelId }}",\n{{ inputs.asTsString }}\n});\n\nconsole.log(chatCompletion.choices[0].message);',
+      "conversationalStream": 'import { OpenAI } from "openai";\n\nconst client = new OpenAI({\n	baseURL: "{{ baseUrl }}",\n	apiKey: "{{ accessToken }}",\n{% if billTo %}\n    defaultHeaders: {\n		"X-HF-Bill-To": "{{ billTo }}" \n	}\n{% endif %}\n});\n\nconst stream = await client.chat.completions.create({\n    model: "{{ providerModelId }}",\n{{ inputs.asTsString }}\n    stream: true,\n});\n\nfor await (const chunk of stream) {\n    process.stdout.write(chunk.choices[0]?.delta?.content || "");\n}'
     }
   },
   "python": {
@@ -2127,13 +2221,13 @@ const image = await client.textToVideo({
       "conversationalStream": 'stream = client.chat.completions.create(\n    model="{{ model.id }}",\n{{ inputs.asPythonString }}\n    stream=True,\n)\n\nfor chunk in stream:\n    print(chunk.choices[0].delta.content, end="") ',
       "documentQuestionAnswering": 'output = client.document_question_answering(\n    "{{ inputs.asObj.image }}",\n    question="{{ inputs.asObj.question }}",\n    model="{{ model.id }}",\n) ',
       "imageToImage": '# output is a PIL.Image object\nimage = client.image_to_image(\n    "{{ inputs.asObj.inputs }}",\n    prompt="{{ inputs.asObj.parameters.prompt }}",\n    model="{{ model.id }}",\n) ',
-      "importInferenceClient": 'from huggingface_hub import InferenceClient\n\nclient = InferenceClient(\n    provider="{{ provider }}",\n    api_key="{{ accessToken }}",\n)',
+      "importInferenceClient": 'from huggingface_hub import InferenceClient\n\nclient = InferenceClient(\n    provider="{{ provider }}",\n    api_key="{{ accessToken }}",\n{% if billTo %}\n    bill_to="{{ billTo }}",\n{% endif %}\n)',
       "textToImage": '# output is a PIL.Image object\nimage = client.text_to_image(\n    {{ inputs.asObj.inputs }},\n    model="{{ model.id }}",\n) ',
       "textToVideo": 'video = client.text_to_video(\n    {{ inputs.asObj.inputs }},\n    model="{{ model.id }}",\n) '
     },
     "openai": {
-      "conversational": 'from openai import OpenAI\n\nclient = OpenAI(\n    base_url="{{ baseUrl }}",\n    api_key="{{ accessToken }}"\n)\n\ncompletion = client.chat.completions.create(\n    model="{{ providerModelId }}",\n{{ inputs.asPythonString }}\n)\n\nprint(completion.choices[0].message) ',
-      "conversationalStream": 'from openai import OpenAI\n\nclient = OpenAI(\n    base_url="{{ baseUrl }}",\n    api_key="{{ accessToken }}"\n)\n\nstream = client.chat.completions.create(\n    model="{{ providerModelId }}",\n{{ inputs.asPythonString }}\n    stream=True,\n)\n\nfor chunk in stream:\n    print(chunk.choices[0].delta.content, end="")'
+      "conversational": 'from openai import OpenAI\n\nclient = OpenAI(\n    base_url="{{ baseUrl }}",\n    api_key="{{ accessToken }}",\n{% if billTo %}\n    default_headers={\n        "X-HF-Bill-To": "{{ billTo }}"\n    }\n{% endif %}\n)\n\ncompletion = client.chat.completions.create(\n    model="{{ providerModelId }}",\n{{ inputs.asPythonString }}\n)\n\nprint(completion.choices[0].message) ',
+      "conversationalStream": 'from openai import OpenAI\n\nclient = OpenAI(\n    base_url="{{ baseUrl }}",\n    api_key="{{ accessToken }}",\n{% if billTo %}\n    default_headers={\n        "X-HF-Bill-To": "{{ billTo }}"\n    }\n{% endif %}\n)\n\nstream = client.chat.completions.create(\n    model="{{ providerModelId }}",\n{{ inputs.asPythonString }}\n    stream=True,\n)\n\nfor chunk in stream:\n    print(chunk.choices[0].delta.content, end="")'
     },
     "requests": {
       "basic": 'def query(payload):\n    response = requests.post(API_URL, headers=headers, json=payload)\n    return response.json()\n\noutput = query({\n    "inputs": {{ providerInputs.asObj.inputs }},\n}) ',
@@ -2143,7 +2237,7 @@ const image = await client.textToVideo({
       "conversationalStream": 'def query(payload):\n    response = requests.post(API_URL, headers=headers, json=payload, stream=True)\n    for line in response.iter_lines():\n        if not line.startswith(b"data:"):\n            continue\n        if line.strip() == b"data: [DONE]":\n            return\n        yield json.loads(line.decode("utf-8").lstrip("data:").rstrip("/n"))\n\nchunks = query({\n{{ providerInputs.asJsonString }},\n    "stream": True,\n})\n\nfor chunk in chunks:\n    print(chunk["choices"][0]["delta"]["content"], end="")',
       "documentQuestionAnswering": 'def query(payload):\n    with open(payload["image"], "rb") as f:\n        img = f.read()\n        payload["image"] = base64.b64encode(img).decode("utf-8")\n    response = requests.post(API_URL, headers=headers, json=payload)\n    return response.json()\n\noutput = query({\n    "inputs": {\n        "image": "{{ inputs.asObj.image }}",\n        "question": "{{ inputs.asObj.question }}",\n    },\n}) ',
       "imageToImage": 'def query(payload):\n    with open(payload["inputs"], "rb") as f:\n        img = f.read()\n        payload["inputs"] = base64.b64encode(img).decode("utf-8")\n    response = requests.post(API_URL, headers=headers, json=payload)\n    return response.content\n\nimage_bytes = query({\n{{ providerInputs.asJsonString }}\n})\n\n# You can access the image with PIL.Image for example\nimport io\nfrom PIL import Image\nimage = Image.open(io.BytesIO(image_bytes)) ',
-      "importRequests": '{% if importBase64 %}\nimport base64\n{% endif %}\n{% if importJson %}\nimport json\n{% endif %}\nimport requests\n\nAPI_URL = "{{ fullUrl }}"\nheaders = {"Authorization": "{{ authorizationHeader }}"}',
+      "importRequests": '{% if importBase64 %}\nimport base64\n{% endif %}\n{% if importJson %}\nimport json\n{% endif %}\nimport requests\n\nAPI_URL = "{{ fullUrl }}"\nheaders = {\n    "Authorization": "{{ authorizationHeader }}",\n{% if billTo %}\n    "X-HF-Bill-To": "{{ billTo }}"\n{% endif %}\n}',
       "tabular": 'def query(payload):\n    response = requests.post(API_URL, headers=headers, json=payload)\n    return response.content\n\nresponse = query({\n    "inputs": {\n        "data": {{ providerInputs.asObj.inputs }}\n    },\n}) ',
       "textToAudio": '{% if model.library_name == "transformers" %}\ndef query(payload):\n    response = requests.post(API_URL, headers=headers, json=payload)\n    return response.content\n\naudio_bytes = query({\n    "inputs": {{ providerInputs.asObj.inputs }},\n})\n# You can access the audio with IPython.display for example\nfrom IPython.display import Audio\nAudio(audio_bytes)\n{% else %}\ndef query(payload):\n    response = requests.post(API_URL, headers=headers, json=payload)\n    return response.json()\n\naudio, sampling_rate = query({\n    "inputs": {{ providerInputs.asObj.inputs }},\n})\n# You can access the audio with IPython.display for example\nfrom IPython.display import Audio\nAudio(audio, rate=sampling_rate)\n{% endif %} ',
       "textToImage": '{% if provider == "hf-inference" %}\ndef query(payload):\n    response = requests.post(API_URL, headers=headers, json=payload)\n    return response.content\n\nimage_bytes = query({\n    "inputs": {{ providerInputs.asObj.inputs }},\n})\n\n# You can access the image with PIL.Image for example\nimport io\nfrom PIL import Image\nimage = Image.open(io.BytesIO(image_bytes))\n{% endif %}',
@@ -2153,12 +2247,15 @@ const image = await client.textToVideo({
   },
   "sh": {
     "curl": {
-      "basic": "curl {{ fullUrl }} \\\n    -X POST \\\n    -H 'Authorization: {{ authorizationHeader }}' \\\n    -H 'Content-Type: application/json' \\\n    -d '{\n{{ providerInputs.asCurlString }}\n    }'",
-      "basicAudio": "curl {{ fullUrl }} \\\n    -X POST \\\n    -H 'Authorization: {{ authorizationHeader }}' \\\n    -H 'Content-Type: audio/flac' \\\n    --data-binary @{{ providerInputs.asObj.inputs }}",
-      "basicImage": "curl {{ fullUrl }} \\\n    -X POST \\\n    -H 'Authorization: {{ authorizationHeader }}' \\\n    -H 'Content-Type: image/jpeg' \\\n    --data-binary @{{ providerInputs.asObj.inputs }}",
+      "basic": "curl {{ fullUrl }} \\\n    -X POST \\\n    -H 'Authorization: {{ authorizationHeader }}' \\\n    -H 'Content-Type: application/json' \\\n{% if billTo %}\n    -H 'X-HF-Bill-To: {{ billTo }}' \\\n{% endif %}\n    -d '{\n{{ providerInputs.asCurlString }}\n    }'",
+      "basicAudio": "curl {{ fullUrl }} \\\n    -X POST \\\n    -H 'Authorization: {{ authorizationHeader }}' \\\n    -H 'Content-Type: audio/flac' \\\n{% if billTo %}\n    -H 'X-HF-Bill-To: {{ billTo }}' \\\n{% endif %}\n    --data-binary @{{ providerInputs.asObj.inputs }}",
+      "basicImage": "curl {{ fullUrl }} \\\n    -X POST \\\n    -H 'Authorization: {{ authorizationHeader }}' \\\n    -H 'Content-Type: image/jpeg' \\\n{% if billTo %}\n    -H 'X-HF-Bill-To: {{ billTo }}' \\\n{% endif %}\n    --data-binary @{{ providerInputs.asObj.inputs }}",
       "conversational": `curl {{ fullUrl }} \\
     -H 'Authorization: {{ authorizationHeader }}' \\
     -H 'Content-Type: application/json' \\
+{% if billTo %}
+    -H 'X-HF-Bill-To: {{ billTo }}' \\
+{% endif %}
     -d '{
 {{ providerInputs.asCurlString }},
         "stream": false
@@ -2166,6 +2263,9 @@ const image = await client.textToVideo({
       "conversationalStream": `curl {{ fullUrl }} \\
     -H 'Authorization: {{ authorizationHeader }}' \\
     -H 'Content-Type: application/json' \\
+{% if billTo %}
+    -H 'X-HF-Bill-To: {{ billTo }}' \\
+{% endif %}
     -d '{
 {{ providerInputs.asCurlString }},
         "stream": true
@@ -2174,7 +2274,10 @@ const image = await client.textToVideo({
     -X POST \\
     -d '{"inputs": {{ providerInputs.asObj.inputs }}, "parameters": {"candidate_labels": ["refund", "legal", "faq"]}}' \\
     -H 'Content-Type: application/json' \\
-    -H 'Authorization: {{ authorizationHeader }}'`
+    -H 'Authorization: {{ authorizationHeader }}'
+{% if billTo %} \\
+    -H 'X-HF-Bill-To: {{ billTo }}'
+{% endif %}`
     }
   }
 };
@@ -2243,23 +2346,34 @@ var HF_JS_METHODS = {
   translation: "translation"
 };
 var snippetGenerator = (templateName, inputPreparationFn) => {
-  return (model, accessToken, provider, providerModelId, opts) => {
+  return (model, accessToken, provider, inferenceProviderMapping, opts) => {
+    const providerModelId = inferenceProviderMapping?.providerId ?? model.id;
     let task = model.pipeline_tag;
     if (model.pipeline_tag && ["text-generation", "image-text-to-text"].includes(model.pipeline_tag) && model.tags.includes("conversational")) {
       templateName = opts?.streaming ? "conversationalStream" : "conversational";
       inputPreparationFn = prepareConversationalInput;
       task = "conversational";
     }
+    let providerHelper;
+    try {
+      providerHelper = getProviderHelper(provider, task);
+    } catch (e) {
+      console.error(`Failed to get provider helper for ${provider} (${task})`, e);
+      return [];
+    }
     const inputs = inputPreparationFn ? inputPreparationFn(model, opts) : { inputs: (0, import_tasks.getModelInputSnippet)(model) };
     const request2 = makeRequestOptionsFromResolvedModel(
-      providerModelId ?? model.id,
+      providerModelId,
+      providerHelper,
       {
         accessToken,
         provider,
         ...inputs
       },
+      inferenceProviderMapping,
       {
-        task
+        task,
+        billTo: opts?.billTo
       }
     );
     let providerInputs = inputs;
@@ -2292,7 +2406,8 @@ var snippetGenerator = (templateName, inputPreparationFn) => {
       },
       model,
       provider,
-      providerModelId: providerModelId ?? model.id
+      providerModelId: providerModelId ?? model.id,
+      billTo: opts?.billTo
     };
     return import_tasks.inferenceSnippetLanguages.map((language) => {
       return CLIENTS[language].map((client) => {
@@ -2382,8 +2497,8 @@ var snippets = {
   "zero-shot-classification": snippetGenerator("zeroShotClassification"),
   "zero-shot-image-classification": snippetGenerator("zeroShotImageClassification")
 };
-function getInferenceSnippets(model, accessToken, provider, providerModelId, opts) {
-  return model.pipeline_tag && model.pipeline_tag in snippets ? snippets[model.pipeline_tag]?.(model, accessToken, provider, providerModelId, opts) ?? [] : [];
+function getInferenceSnippets(model, accessToken, provider, inferenceProviderMapping, opts) {
+  return model.pipeline_tag && model.pipeline_tag in snippets ? snippets[model.pipeline_tag]?.(model, accessToken, provider, inferenceProviderMapping, opts) ?? [] : [];
 }
 function formatBody(obj, format) {
   switch (format) {