Updating to AI SDK 6.2.3

This update compares two baselines: the application SDK moves from 6.0.0 to 6.2.3, while the published documentation last listed the 5.24.0 catalog. The current production catalog contains 201 models across 31 providers: 66 image, 84 video, 24 audio, and 27 text.

SDK changes since 6.0.0

The SDK adds 11 production text-model IDs with no production model removals: claude-sonnet-4-5, claude-sonnet-5, gemini-2.5-flash, gpt-4.1-mini, gpt-4.1-nano, gpt-4o, gpt-4o-mini, gpt-5, gpt-5-mini, gpt-5.1, and gpt-5.2. These IDs are available through the typed text-generation API and model constants.

Result metadata migration

Version 6.1.0 removes seed, nsfw, width, height, contentType, duration, fps, and fileSize from items[].metadata, including runtime promotion. Update consumers that read these properties. exploreImageId and generatedVoiceId remain supported.

The new optional lastFrameUrl promotes a vendor's last_frame_url for frame chaining. Request the last frame from a supporting Seedance model, then use the returned URL as a subsequent start frame.

const first = await client.generate('seedance-2.5', {
  prompt: 'A paper boat drifts down a quiet stream',
  returnLastFrame: true,
});
const startFrame = first.items[0]?.metadata?.lastFrameUrl;
if (startFrame) {
  await client.generate('seedance-2.5', {
    prompt: 'The boat passes under a stone bridge',
    startFrame,
  });
}

The URL is optional because it depends on the vendor response. Check it before starting the next generation.

Seedance output compatibility

Version 6.2.3 adds colorDepth: '10bit' | '8bit' alongside the existing outputFormat: 'mp4' | 'mov' container setting. MP4 and MOV support were already present in 6.0.0. Color depth is selectable for 1080p MP4 only; 480p and 720p already use 8-bit, while MOV does not offer an 8-bit variant. The default remains 10-bit.

const video = await client.generate('seedance-2.5', {
  prompt: 'A bird lands on a garden fence',
  resolution: '1080p',
  outputFormat: 'mp4',
  colorDepth: '8bit',
});

This requests the worker's H.264 8-bit re-encode (mp4_8bit) for compatibility with browsers and editors. The same setting applies to the 2.5 edit and extend variants. The SDK also exposes preview variants, which are excluded from this site's production catalog.

Node.js 20 or newer is still required. The workflow types dependency advances from ^1.1.137 to ^1.1.139; package exports stay unchanged. See the official source comparison.

Changes since the published documentation

The published catalog grows from 181 to 201 entries: 26 added and 6 removed. Meta joins the active provider list; Reve leaves it, keeping the provider total unchanged. Historical provider links remain available.

Added catalog entries

Model ID	Name	Mode
claude-fable-5	Claude Fable 5	text
claude-fable-5-1	Claude Fable 5.1	text
claude-opus-5	Claude Opus 5	text
claude-sonnet-4-5	Claude Sonnet 4.5	text
claude-sonnet-5	Claude Sonnet 5	text
flux-video-edit	FLUX Video Edit	video
gemini-2.5-flash	Gemini 2.5 Flash	text
gemini-3.8-flash	Gemini 3.8 Flash	text
gpt-4.1-mini	GPT-4.1 Mini	text
gpt-4.1-nano	GPT-4.1 Nano	text
gpt-4o	GPT-4o	text
gpt-4o-mini	GPT-4o Mini	text
gpt-5	GPT-5	text
gpt-5-mini	GPT-5 Mini	text
gpt-5.1	GPT-5.1	text
gpt-5.2	GPT-5.2	text
gpt-5.6-luna	GPT-5.6 Luna	text
gpt-5.6-sol	GPT-5.6 Sol	text
gpt-5.6-terra	GPT-5.6 Terra	text
gpt-6-astra	GPT-6 Astra	text
gpt-image-2.5-flare	GPT Image 2.5 Flare	image
gpt-image-2.5-sunburst	GPT Image 2.5 Sunburst	image
lyria-3.5	Lyria 3.5	audio
minimax-h3-max-camera-controls	MiniMax H3 Max Camera Controls	video
minimax-h3-max-turbo	MiniMax H3 Max Turbo	video
muse-image-1.0	Muse Image 1.0	image

Removed from the production catalog

picsart-qwen-image-edit-angle, qwen-image-2, reve, sora-2, sora-2-extend, sora-2-pro. These IDs are no longer listed by the SDK's production catalog; this does not imply that a vendor discontinued the underlying model.

Provider tables and model parameter sections reflect the current catalog. For current discovery, use the Model Catalog or gen-ai models info <id> --json.