Open History to see the last commit
1---
2title: BytePlus
3id: byteplus-adapter
4order: 8
5description: "Use BytePlus ModelArk with TanStack AI — Seed chat models with reasoning, Seedance video generation, Seedream image generation, and Seed Speech text-to-speech and transcription via @tanstack/ai-byteplus."
6keywords:
7 - tanstack ai
8 - byteplus
9 - modelark
10 - ark
11 - seed
12 - seedance
13 - seedream
14 - seed speech
15 - bytedance
16 - video generation
17 - image generation
18 - text to speech
19 - transcription
20 - adapter
21---
22
23The BytePlus adapter connects TanStack AI to [BytePlus](https://www.byteplus.com/), ByteDance's international model platform, across four generation kinds:
24
25- **Chat** — the Seed model family (plus GLM, DeepSeek and gpt-oss) on ModelArk's OpenAI-compatible `/chat/completions` endpoint, with reasoning on by default.
26- **Video** — Seedance, an asynchronous task API.
27- **Image** — Seedream.
28- **Speech** — Seed Speech text-to-speech and transcription, which live on a **different host with a different API key** (see [Two products, two keys](#two-products-two-keys)).
29
30## Installation
31
32```bash
33npm install @tanstack/ai-byteplus
34# or
35pnpm add @tanstack/ai-byteplus
36# or
37yarn add @tanstack/ai-byteplus
38```
39
40## Two products, two keys
41
42BytePlus splits these models across two products, and they do not share credentials:
43
44| Adapters | Product | Env var | Auth header |
45| --- | --- | --- | --- |
46| `byteplusText`, `byteplusVideo`, `byteplusImage` | ModelArk (Ark) | `ARK_API_KEY` (falls back to `BYTEPLUS_API_KEY`) | `Authorization: Bearer` |
47| `byteplusSpeech`, `byteplusTranscription` | Seed Speech | `BYTEPLUS_VOICE_API_KEY` | `X-Api-Key` |
48
49```bash
50# ModelArk: chat, Seedance video, Seedream image
51ARK_API_KEY=...
52
53# Seed Speech: TTS and transcription — a separate product key
54BYTEPLUS_VOICE_API_KEY=...
55```
56
57Passing an Ark key to the speech adapters fails with `45000010 Invalid X-Api-Key`.
58
59Ark keys are also **region-isolated**: the default base URL is the Asia-Pacific south-east endpoint (`https://ark.ap-southeast.bytepluses.com/api/v3`), and a key issued for one region will not authenticate against another. Point the adapter elsewhere with `baseURL`:
60
61```typescript
62import { createBytePlusText } from '@tanstack/ai-byteplus'
63import { arkApiKey } from './config'
64
65const adapter = createBytePlusText('dola-seed-2-1-turbo-260628', arkApiKey, {
66 baseURL: 'https://ark.eu-west.bytepluses.com/api/v3',
67})
68```
69
70Per the BytePlus docs the EU endpoint serves chat and image only — Seedance video is Asia-Pacific only.
71
72## Chat
73
74The adapter carries the model, so there is no separate `model` option.
75
76**Server** — an endpoint that streams the reply back over SSE:
77
78```typescript
79import { chat, toServerSentEventsResponse } from '@tanstack/ai'
80import { byteplusText } from '@tanstack/ai-byteplus'
81
82export async function POST(request: Request) {
83 const { messages } = await request.json()
84
85 const stream = chat({
86 adapter: byteplusText('dola-seed-2-1-turbo-260628'),
87 messages,
88 })
89
90 return toServerSentEventsResponse(stream)
91}
92```
93
94**Client** — the same `useChat` hook as every other provider:
95
96```tsx
97import { useState } from 'react'
98import { fetchServerSentEvents, useChat } from '@tanstack/ai-react'
99
100export function Chat() {
101 const [input, setInput] = useState('')
102
103 const { messages, sendMessage, isLoading } = useChat({
104 connection: fetchServerSentEvents('/api/chat'),
105 })
106
107 return (
108 <div>
109 {messages.map((message) => (
110 <div key={message.id}>
111 <strong>{message.role}</strong>
112 {message.parts.map((part, index) =>
113 part.type === 'text' ? <p key={index}>{part.content}</p> : null,
114 )}
115 </div>
116 ))}
117
118 <form
119 onSubmit={(event) => {
120 event.preventDefault()
121 if (!input.trim() || isLoading) return
122 sendMessage(input)
123 setInput('')
124 }}
125 >
126 <input value={input} onChange={(e) => setInput(e.target.value)} />
127 <button type="submit" disabled={isLoading}>
128 Send
129 </button>
130 </form>
131 </div>
132 )
133}
134```
135
136### Model options
137
138Ark's chat endpoint is OpenAI-compatible, so sampling parameters keep their OpenAI snake_case names and live in `modelOptions`. `thinking`, `reasoning_effort`, `repetition_penalty` and `service_tier` are the Ark-only additions:
139
140```typescript
141import { chat, toServerSentEventsResponse } from '@tanstack/ai'
142import { byteplusText } from '@tanstack/ai-byteplus'
143
144export async function POST(request: Request) {
145 const { messages } = await request.json()
146
147 const stream = chat({
148 adapter: byteplusText('dola-seed-2-1-turbo-260628'),
149 messages,
150 modelOptions: {
151 temperature: 0.7,
152 top_p: 0.9,
153 max_tokens: 2048,
154 thinking: { type: 'enabled' },
155 reasoning_effort: 'medium',
156 },
157 })
158
159 return toServerSentEventsResponse(stream)
160}
161```
162
163Two constraints the type system can't express, both live-verified as `400`s:
164
165- `max_tokens` and `max_completion_tokens` are mutually exclusive.
166- `reasoning_effort` cannot be combined with `thinking: { type: 'disabled' }`.
167
168`service_tier: 'flex'` routes the request to the cheaper offline batch queue with no latency guarantee.
169
170## Reasoning and `encrypted_content`
171
172Seed models reason by default. Reasoning arrives as its own stream of `reasoning_content` deltas and is surfaced as reasoning content rather than answer text, so `useChat` renders it separately from the reply. Turn it off per request:
173
174```typescript
175import { chat, toServerSentEventsResponse } from '@tanstack/ai'
176import { byteplusText } from '@tanstack/ai-byteplus'
177
178export async function POST(request: Request) {
179 const { messages } = await request.json()
180
181 const stream = chat({
182 adapter: byteplusText('dola-seed-2-1-turbo-260628'),
183 messages,
184 modelOptions: { thinking: { type: 'disabled' } },
185 })
186
187 return toServerSentEventsResponse(stream)
188}
189```
190
191`disabled` works everywhere; `auto` is accepted only by `gpt-oss-120b-250805`. `deepseek-v3-2-251201` is the one model that defaults to reasoning *off*.
192
193The four "thinking summary" models — `dola-seed-2-1-turbo-260628`, `seed-2-0-lite-260428`, `seed-2-0-mini-260428` and `seed-2-0-pro-260328` — also emit an opaque `encrypted_content` blob alongside the reasoning trace. It is a signature over that trace, and BytePlus's docs ask for it back verbatim on the assistant message in the next turn.
194
195**The adapter round-trips it for you**, over the same seam Anthropic's thinking signatures use: the blob is captured off the stream and attached to the reasoning step as its `signature`, which lands on the thinking part of the assistant message and is echoed back on the next request. Two consequences worth knowing:
196
197- If you persist and replay conversation history yourself, keep the thinking parts' `signature` — dropping it costs you the reasoning-cache hit. It is not fatal: a live probe confirmed Ark accepts a turn whose assistant message omits `encrypted_content`.
198- A structured-output turn doesn't capture the blob, for the same reason — again a lost cache hit, not a failed request.
199
200## Structured output
201
202Ten of the eighteen chat models accept `response_format: { type: 'json_schema' }`:
203
204`dola-seed-2-1-turbo-260628`, `seed-2-0-pro-260328`, `seed-2-0-lite-260228`, `seed-2-0-mini-260215`, `seed-1-8-251228`, `seed-1-6-250915`, `seed-1-6-250615`, `seed-1-6-flash-250715`, `seed-1-6-flash-250615`, `glm-5-2-260617`.
205
206```typescript
207import { chat } from '@tanstack/ai'
208import { byteplusText } from '@tanstack/ai-byteplus'
209import { z } from 'zod'
210
211const RecipeSchema = z.object({
212 name: z.string().meta({ description: 'Name of the dish' }),
213 minutes: z.number().meta({ description: 'Total cooking time in minutes' }),
214 ingredients: z.array(z.string()),
215})
216
217const recipe = await chat({
218 adapter: byteplusText('dola-seed-2-1-turbo-260628'),
219 messages: [{ role: 'user', content: 'Give me a recipe for carbonara' }],
220 outputSchema: RecipeSchema,
221})
222
223console.log(recipe.name, recipe.minutes)
224```
225
226On the other eight models the adapter **fails loudly** rather than degrading: `chat({ outputSchema })` throws, and the streaming form emits a `RUN_ERROR` naming the models that do work. There is no JSON-mode fallback to fall back *to* — Ark rejects `response_format: { type: 'json_object' }` on every one of them too.
227
228Two traps worth knowing, both found by probing the live API rather than reading the tables:
229
230- **The published capability tables are wrong in both directions.** `seed-2-0-lite-260428` — the obvious default model — is one of the eight that reject a JSON schema. Reach for `seed-2-0-lite-260228` or `dola-seed-2-1-turbo-260628` when you need typed output.
231- **`glm-4-7-251222` accepts a schema and then ignores it**, answering in prose with a `200`. It is deliberately excluded from the supported list, so the adapter rejects it up front instead of handing you unparseable output.
232
233`BYTEPLUS_STRUCTURED_OUTPUT_CHAT_MODELS` is exported if you want to gate a model picker on it.
234
235## Video generation (Seedance)
236
237> Video generation is an [experimental feature](../media/video-generation).
238
239Seedance is an asynchronous task API: `generateVideo()` opens the job and returns a `jobId`, and the video URL arrives with the terminal status.
240
241```typescript
242import { generateVideo, getVideoJobStatus } from '@tanstack/ai'
243import { byteplusVideo } from '@tanstack/ai-byteplus'
244
245const adapter = byteplusVideo('dreamina-seedance-2-0-260128')
246
247const { jobId } = await generateVideo({
248 adapter,
249 prompt: 'a guitar being played in a store',
250 size: '16:9_720p',
251 duration: 5,
252})
253
254let status = await getVideoJobStatus({ adapter, jobId })
255while (status.status === 'pending' || status.status === 'processing') {
256 await new Promise((resolve) => setTimeout(resolve, 5000))
257 status = await getVideoJobStatus({ adapter, jobId })
258}
259
260console.log(status.status === 'completed' ? status.url : status.error)
261```
262
263**Generated video URLs expire 24 hours after the task completes** (the task record itself is kept for seven days), so download anything you intend to keep — see [Keeping generated files](../persistence/keep-generated-files).
264
265`size` is a `ratio` or `ratio_resolution` template (`'16:9'`, `'16:9_720p'`). Ratios are `16:9`, `9:16`, `4:3`, `3:4`, `1:1`, `21:9` and `adaptive`.
266
267### Driving the whole lifecycle from the client
268
269**Server** — hand the polling to the core with `stream: true` and pipe the chunks out:
270
271```typescript
272import { generateVideo, toServerSentEventsResponse } from '@tanstack/ai'
273import { byteplusVideo } from '@tanstack/ai-byteplus'
274
275export async function POST(request: Request) {
276 const { prompt } = await request.json()
277
278 const stream = generateVideo({
279 adapter: byteplusVideo('dreamina-seedance-2-0-260128'),
280 prompt,
281 size: '16:9_720p',
282 duration: 5,
283 stream: true,
284 pollingInterval: 5000,
285 })
286
287 return toServerSentEventsResponse(stream)
288}
289```
290
291**Client** — `useGenerateVideo` tracks the job for you:
292
293```tsx
294import { fetchServerSentEvents, useGenerateVideo } from '@tanstack/ai-react'
295
296export function SeedanceGenerator() {
297 const { generate, result, videoStatus, isLoading, error } = useGenerateVideo({
298 connection: fetchServerSentEvents('/api/generate/video'),
299 })
300
301 return (
302 <div>
303 <button
304 onClick={() => generate({ prompt: 'a guitar being played in a store' })}
305 disabled={isLoading}
306 >
307 {isLoading ? 'Generating…' : 'Generate video'}
308 </button>
309
310 {isLoading && <p>Status: {videoStatus?.status ?? 'starting…'}</p>}
311 {error && <p>Error: {error.message}</p>}
312 {result && <video src={result.url} controls width={640} />}
313 </div>
314 )
315}
316```
317
318### Per-model options
319
320Seedance options are model-specific, and **Ark rejects an inapplicable field with a `400` rather than ignoring it** — "the specified parameter `draft` is not supported for model seedance-1-0-pro in t2v, must be empty". The applicable set:
321
322| Option | Models that accept it |
323| --- | --- |
324| `service_tier` (`'default'` \| `'flex'`) | Seedance 1.x only — Seedance 2.5 and the 2.0 family reject it |
325| `camera_fixed` | Seedance 1.x only |
326| `frames` (fractional-second output, `25 + 4n` in `[29, 289]`) | `seedance-1-0-pro-250528`, `seedance-1-0-pro-fast-251015` |
327| `draft` (cheap low-fidelity preview) | `seedance-1-5-pro-251215` only |
328| `priority` (`0`–`9`) | Seedance 2.5 and the `dreamina-seedance-2-0-*` family |
329| `duration: -1` (model picks the length) | Seedance 2.5, 2.0 and `seedance-1-5-pro-251215` |
330| `output_format` (`'mp4'` \| `'mov'`) | Seedance 2.5 (default `mp4`) |
331| `seed`, `watermark`, `generate_audio`, `return_last_frame`, `callback_url` | every model |
332
333`watermark` defaults to `false` for video (the opposite of Seedream images). `generate_audio` is accepted everywhere but only Seedance 2.5, 2.0 and 1.5-pro actually produce an audio track.
334
335Resolutions are per model too: there is **no 2K tier on any Seedance model**, `4k` exists only on `dreamina-seedance-2-0-260128`, Seedance 2.5 is **480p/720p only**, and `seedance-1-0-pro-fast-251015` does accept `1080p` despite older prose listing it as 480p/720p only. An unsupported combination is caught locally with a clear error before the request goes out.
336
337Reference media follows the shared [role hints](../media/video-generation#role-hints) — `start_frame`, `end_frame` and `reference`. Seedance 2.5 and the 2.0 family take full multimodal references (reference images, video and audio); Seedance 2.5 additionally accepts **audio-only** reference input and up to 30 reference images / 10 videos / 10 audio clips (2.0 is 9 / 3 / 3). The 1.x models take start/end frames only, and `seedance-1-0-pro-fast-251015` takes a start frame only.
338
339### Seedance 2.5
340
341Seedance 2.5 (`dreamina-seedance-2-5-260628`) is a first-class model in this package. It is the current multimodal flagship for longer clips (up to **30 seconds**), audio-only reference input, and `output_format: 'mov'`. Like the 2.0 series it may still require model activation / a resource pack in the Ark Console before the account can call it (`404 ModelNotOpen` until then).
342
343```typescript
344import { generateVideo } from '@tanstack/ai'
345import { byteplusVideo } from '@tanstack/ai-byteplus'
346
347const { jobId } = await generateVideo({
348 adapter: byteplusVideo('dreamina-seedance-2-5-260628'),
349 prompt: 'a guitar being played in a store',
350 size: '16:9_720p',
351 duration: 10,
352 modelOptions: {
353 generate_audio: true,
354 priority: 5,
355 output_format: 'mp4',
356 },
357})
358```
359
360| Capability | Seedance 2.5 |
361| --- | --- |
362| Duration | 4–30s, or `-1` (model chooses; required for video-editing tasks) |
363| Resolution | `480p`, `720p` (default `720p`) — no 1080p / 4k |
364| Reference media | images 1–30, videos 0–10, audio 0–10; **audio-only allowed** |
365| First + last frame | yes |
366| `priority` / `generate_audio` | yes |
367| `service_tier` / `camera_fixed` / `frames` / `draft` | no |
368
369`byteplusVideo()` still accepts any string for ids BytePlus ships between package releases. An unknown id relaxes local guards so Ark validates the request; known ids (including 2.5) keep their documented narrowing.
370
371### Seedance here vs. Seedance via fal
372
373Seedance is also reachable through [`@tanstack/ai-fal`](./fal), which proxies it alongside hundreds of other hosted models. This adapter talks to BytePlus directly, which means BytePlus billing and rate limits, model ids in BytePlus's own naming, and the first-class Seedance request fields above (`camera_fixed`, `draft`, `priority`, reference-media roles) rather than fal's normalized subset. Pick whichever matches the account you already have; there is no reason to install both just for Seedance.
374
375## Image generation (Seedream)
376
377```typescript
378import { generateImage } from '@tanstack/ai'
379import { byteplusImage } from '@tanstack/ai-byteplus'
380
381const result = await generateImage({
382 adapter: byteplusImage('dola-seedream-5-0-pro-260628'),
383 prompt: 'a guitar being played in a store',
384 size: '2K',
385 modelOptions: { watermark: false },
386})
387
388console.log(result.images[0]?.url)
389```
390
391`size` takes either a token (`1K`, `2K`, `4K`) or explicit pixels (`2048x2048`) — never a mix of the two. Pass image parts in the `prompt` array to edit or condition on existing images (up to 14 references, 10 on `dola-seedream-5-0-pro-260628`):
392
393```typescript
394import { generateImage } from '@tanstack/ai'
395import { byteplusImage } from '@tanstack/ai-byteplus'
396
397const result = await generateImage({
398 adapter: byteplusImage('seedream-5-0-260128'),
399 prompt: [
400 { type: 'text', content: 'Put this guitar on a concert stage' },
401 {
402 type: 'image',
403 source: { type: 'url', value: 'https://example.com/guitar.png' },
404 },
405 ],
406})
407
408console.log(result.images.length)
409```
410
411Two behaviours surprise people:
412
413- **`watermark` defaults to `true`.** BytePlus stamps "AI generated" into the bottom-right corner unless you pass `watermark: false`. The adapter never sets the field implicitly, so the provider default applies.
414- **`numberOfImages` is an upper bound, not a count.** Seedream has no `n` parameter; asking for more than one image maps onto its group-image mode (`sequential_image_generation: 'auto'` with `max_images`), where the model decides how many images the prompt actually warrants. A request for four can come back with two, and the adapter logs a warning when it does.
415
416Result URLs expire after 24 hours. Pass `response_format: 'b64_json'` in `modelOptions` to get the bytes inline instead.
417
418## Text-to-speech (Seed Speech)
419
420Seed Speech uses `BYTEPLUS_VOICE_API_KEY`, not the Ark key.
421
422```typescript
423import { generateSpeech } from '@tanstack/ai'
424import { byteplusSpeech } from '@tanstack/ai-byteplus'
425
426const result = await generateSpeech({
427 adapter: byteplusSpeech('seed-audio-1.0'),
428 text: 'welcome to the guitar store',
429 voice: 'en_female_stokie_uranus_bigtts',
430 format: 'mp3',
431})
432
433console.log(result.contentType, result.audio.length)
434```
435
436Seed Speech has **no top-level `speaker` field** — the voice travels inside a
437`references` entry, so the adapter turns `voice` into
438`references: [{ speaker: 'en_female_stokie_uranus_bigtts' }]` for you. The
439consequence: `modelOptions.references` **replaces** that entry rather than
440merging with it, so a request that passes `references` for voice cloning
441silently drops `voice`. Include a `speaker` member yourself if you still want a
442stock voice alongside your clips.
443
444The suffix on a voice id tells you which generation you are asking for:
445
446- `_uranus_bigtts` — TTS 2.0 voices, the current generation.
447- `_mars_bigtts` / `_moon_bigtts` — TTS 1.0 voices.
448- `*_emo_v2_*` — TTS 1.0 voices that additionally accept emotion tags.
449
450The full roster lives in the [BytePlus voice list](https://docs.byteplus.com/en/docs/byteplusvoice/voicelist) and changes far more often than this package ships, so any string is accepted.
451
452Output formats are `wav`, `mp3`, `pcm` and `ogg_opus`. Reach for `modelOptions` for `ogg_opus`, for an explicit `sample_rate`, for `references` (voice cloning — up to three 30-second audio clips, addressed from the text as `@Audio1`–`@Audio3`), for `watermark`, or for word-level timings:
453
454```typescript
455import { generateSpeech } from '@tanstack/ai'
456import { byteplusSpeech, type BytePlusTTSResult } from '@tanstack/ai-byteplus'
457
458const result: BytePlusTTSResult = await generateSpeech({
459 adapter: byteplusSpeech('seed-audio-1.0'),
460 text: 'welcome to the guitar store',
461 modelOptions: {
462 format: 'ogg_opus',
463 sample_rate: 48000,
464 speech_rate: 20,
465 enable_subtitle: true,
466 },
467})
468
469for (const sentence of result.subtitle?.sentences ?? []) {
470 console.log(sentence.text, sentence.start_time)
471}
472```
473
474Three things to plan around:
475
476- **Synthesis is capped at 120 seconds of output.** The cap applies to `originalDuration` (the length before `speech_rate` is applied), which is also what BytePlus bills on.
477- **Subtitle timings are milliseconds** while `duration` and `originalDuration` are seconds. The units genuinely differ.
478- The `url` on the result expires roughly two hours after generation — persist the base64 `audio`, not the link.
479
480The client half is the standard [`useGenerateSpeech` hook](../media/text-to-speech#streaming-mode-server-route--client-hook); nothing about it is BytePlus-specific.
481
482## Transcription (Seed Speech ASR)
483
484Also on the voice key. The endpoint is synchronous — audio in, transcript out, no polling:
485
486```typescript
487import { generateTranscription } from '@tanstack/ai'
488import { byteplusTranscription } from '@tanstack/ai-byteplus'
489import { audioFile } from './audio'
490
491const result = await generateTranscription({
492 adapter: byteplusTranscription('seed-asr'),
493 audio: audioFile,
494 modelOptions: { enable_punc: true, enable_speaker_info: true },
495})
496
497console.log(result.text)
498for (const segment of result.segments ?? []) {
499 console.log(segment.speaker, segment.text)
500}
501```
502
503Audio can be a `File`, base64, a data URL or a public URL, up to two hours and 100 MB. `enable_itn` renders spoken numbers and dates as digits, `enable_punc` inserts punctuation, `enable_ddc` strips fillers, and `enable_speaker_info` attaches speaker labels that surface as `segment.speaker`.
504
505## Supported models
506
507- **Chat** — `dola-seed-2-1-turbo-260628`, `seed-2-0-lite-260428`, `seed-2-0-mini-260428`, `seed-2-0-pro-260328`, `seed-2-0-lite-260228`, `seed-2-0-mini-260215`, `seed-2-0-code-preview-260328`, `seed-1-8-251228`, `seed-1-6-250915`, `seed-1-6-250615`, `seed-1-6-flash-250715`, `seed-1-6-flash-250615`, `glm-5-2-260617`, `glm-4-7-251222`, `deepseek-v4-pro-260425`, `deepseek-v4-flash-260425`, `deepseek-v3-2-251201`, `gpt-oss-120b-250805`.
508- **Video** — `dreamina-seedance-2-5-260628`, `dreamina-seedance-2-0-260128`, `dreamina-seedance-2-0-fast-260128`, `dreamina-seedance-2-0-mini-260615`, `seedance-1-5-pro-251215`, `seedance-1-0-pro-250528`, `seedance-1-0-pro-fast-251015`.
509- **Image** — `dola-seedream-5-0-pro-260628`, `seedream-5-0-260128`, `seedream-5-0-lite-260128`, `seedream-4-5-251128`, `seedream-4-0-250828`.
510- **Speech** — `seed-audio-1.0` (TTS) and `seed-asr` (transcription).
511
512BytePlus retires model ids aggressively and its published lists include ids that no longer resolve, so this package ships only dated ids that answered a live request. `BYTEPLUS_CHAT_MODELS`, `BYTEPLUS_VIDEO_MODELS`, `BYTEPLUS_IMAGE_MODELS`, `BYTEPLUS_TTS_MODELS` and `BYTEPLUS_TRANSCRIPTION_MODELS` are the authoritative lists and are exported for building model pickers.
513
514## API Reference
515
516Every factory has an environment-variable form and an explicit-key form.
517
518### `byteplusText(model, config?)` / `createBytePlusText(model, apiKey, config?)`
519
520Chat adapter for the Seed, GLM, DeepSeek and gpt-oss models. Reads `ARK_API_KEY` (or `BYTEPLUS_API_KEY`). `config.baseURL` overrides the region endpoint.
521
522### `byteplusVideo(model, config?)` / `createBytePlusVideo(model, apiKey, config?)`
523
524Seedance video adapter (experimental). Reads `ARK_API_KEY`.
525
526### `byteplusImage(model, config?)` / `createBytePlusImage(model, apiKey, config?)`
527
528Seedream image adapter. Reads `ARK_API_KEY`.
529
530### `byteplusSpeech(model, config?)` / `createBytePlusSpeech(model, apiKey, config?)`
531
532Seed Speech TTS adapter. Reads **`BYTEPLUS_VOICE_API_KEY`**.
533
534### `byteplusTranscription(model, config?)` / `createBytePlusTranscription(model, apiKey, config?)`
535
536Seed Speech transcription adapter. Reads **`BYTEPLUS_VOICE_API_KEY`**.
537
538## Provider Tools
539
540BytePlus does not expose provider-specific tool factories. Define your own with `toolDefinition()` from `@tanstack/ai` — Ark's tool calling is the standard OpenAI shape and works with the usual [tools flow](../tools/tools.md).
541
542## Next Steps
543
544- [Video Generation](../media/video-generation) — the full jobs/polling flow and the `useGenerateVideo` hook
545- [Image Generation](../media/image-generation) — image-conditioned generation and role hints
546- [Structured Outputs](../structured-outputs/one-shot) — schemas, validation and streaming
547- [Other Adapters](./openai) — explore other providers