Google Flow API v1— added support for the new Google AI Plus tier (userPaygateTier: PAYGATE_TIER_ZERO, sku: G1_TIER0). Plus accounts route identically to Pro (same SERVICE_TIER_INTERMEDIATE model access). Previously the API rejected Plus accounts with 500 Unknown account paygate tier.

Google Flow API v1 — reusable Voices and Characters. Create custom voices via POST /voices and named characters (1–2 images + optional voice) via POST /characters. Use them in POST /videos and POST /images via character_1..7.

Examples

Google Flow API v1 — omni-flash now supports image-to-video (start frame) via startImage on POST /videos.

Google Flow API v1 — inline @-mention markers in prompt on POST /videos and POST /images. Position your references inside the prompt for tighter compositional control:
• /videos: @character_1..7, @referenceImage_1..7, @referenceAudio_1..5
• /images: @character_1..7, @reference_1..10
• Case-insensitive; opt-in (prompts without markers are unchanged); each marker MUST have a matching body slot supplied or returns 400.

Examples

Google Flow API v1 — load-balancing tuned for fairer distribution across pools of healthy accounts:
• Scoring formula updated to include completed jobs in the rolling 15-minute window, so a heavy-volume account self-limits during its window instead of monopolizing the pool. Formula is now score = executing + completed + (failed × 10) + (rateLimited × 20). See Load Balancing Algorithm for details.
• PUBLIC_ERROR_USER_QUOTA_REACHED quarantine TTL shortened from "until next UTC midnight" to ~30 minutes after the most recent hit — matches Google's actual quota recovery window and avoids 20+ hour over-blocks when the quota clears earlier. PUBLIC_ERROR_PER_MODEL_DAILY_QUOTA_REACHED keeps its until-midnight TTL.

No API surface changes, behavior change only. No action required.

Google Flow API v1 — June 15 update: new asset endpoint, clearer rate-limit errors, and step-by-step guides.

• New endpoint GET /assets/mediaGenerationId — resolve a signed Google Cloud Storage download URL (~6h) for a previously uploaded image or video asset.
• All 429 responses now carry an HTTP Retry-After header + body retryAfter ISO. Cooldowns: USER_QUOTA_REACHED / USER_REQUESTS_THROTTLED = 30 min, PER_MODEL_DAILY_QUOTA_REACHED = next UTC midnight, captcha / unknown = 60s.
• captcha_quality: prefix on the top-level error for low-score or rejected reCAPTCHA 429 / 403; other 429s surface Google's reason directly.
• GET /accounts/email returns a recommendations array when fewer than two captcha providers are configured.
• POST /assets/email now streams uploads to Google (no in-worker memory ceiling on large video; 20 MB image / 100 MB video limits unchanged).
• Breaking change — POST /images and POST /images/upscale 429 / 403 shapes now align with POST /videos: top-level error = classification string, code = HTTP status, Google's error body + captcha move under a nested response. Migrate: body.error.* → body.response.error.*, body.captcha → body.response.captcha.

New step-by-step guides with copy-paste curl and runnable Node.js batch scripts:
• Veo 3.1 video — https://useapi.net/docs/articles/google-flow-bash
• Images (Nano Banana / Nano Banana Pro / Imagen 4) — https://useapi.net/docs/articles/google-flow-images-bash
• Gemini Omni Flash audio-native video — https://useapi.net/docs/articles/omni-flash-bash

Source code: https://github.com/useapi/google-flow-api
Full changelog: https://useapi.net/docs/changelog

Runway API v1 — updates to the Seedance 2.0 family on POST /videos:

• New model seedance-2-fast — a faster, lower-cost variant of seedance-2 (about 20% fewer credits per second of output) that takes exactly the same request parameters. Use it for iteration and drafts; seedance-2 remains the higher-quality option for final renders.
• Audio reference inputs are now supported on both seedance-2 and seedance-2-fast. Pass up to three audio assets via audioAssetId1, audioAssetId2, and audioAssetId3 (.mp3, .wav, or .m4a). Their combined duration must not exceed 15 seconds, and audio references must be combined with at least one image or video reference.
• Requests that combine resolution: "1080p" with exploreMode: true are now rejected with HTTP 400. Per the Runway help docs, 1080p output requires Credits Mode — set exploreMode: false to generate at 1080p.
• The total duration across videoAssetId, videoAssetId2, and videoAssetId3 is now capped at 15 seconds (previously only each video's own duration was checked).
• The reference-image cap has been tightened from 11 to 9. The combined reference cap across images, videos, and audio is also 9.

See Model Comparison: Seedance for the full parameter matrix.

Runway API v1 — new model grok-imagine-1.5 (xAI Grok Imagine 1.5) on POST /videos.
Image-to-video only; pass the first frame via imageAssetId1.
Aspect ratios 2:3 and 3:2 are now accepted (grok-only).
See Model Comparison: Grok for the full matrix.

Kling API v1 — new model kling-v3-0-turbo (Kling 3.0 Turbo), a faster Kling 3.0 variant at full v3.0 quality, on POST /videos/text2video and POST /videos/image2video-frames:
• std/pro modes, 3`–`15s duration, audio always on — put spoken dialogue right in the prompt.
• Built-in multi-shot bias lets one prompt carry scene changes. Does not support 4k, explicit multi-shot, end frame, or element references.
• POST /videos/image2video-frames now also accepts up to 3 element references
(element_1`–`element_3) on Kling v3.0.
• Kling v3.0 now supports Native Audio at 4k resolution.

Examples

PixVerse API v2 — Grok Imagine 1.5, Grok video extend, and Seedance 2.0 4K:

• New model grok-imagine-1.5 (xAI Grok Imagine 1.5) on POST /videos/create. Image-to-video only — pass the first frame via first_frame_path (no text-to-video). Quality 480p / 720p, duration 1-15s, native audio.
• POST /videos/extend now supports grok-imagine (480p / 720p, 2-10s, native audio). The legacy v5 and v5.5 extend modes have been retired, so extend now runs on v6 and grok-imagine.
• seedance-2.0 adds 4K. Pass quality: 2160p on POST /videos/create (price multiplier ×8). seedance-2.0-fast stays at 480p / 720p. 4K is delivered as HEVC.

Examples

Dreamina API v1 — new image and video models:

• New video model seedance-2.0-mini on POST /videos — the most cost-effective Seedance 2.0 (CA region, 720p, 4–15s) with Omni Reference, first/last frame, and text-to-video. Reference uploaded images in the prompt with @imageN.
• seedance-2.0 adds a 4k resolution tier on CA accounts, alongside 720p and 1080p.
• New image model seedream-4.7 (Image 4.7) on POST /images — CA region, 2k/4k, up to 6 reference images.

Examples

Midjourney service notice

Midjourney support on useapi.net has been discontinued..

We're not able to share the reason or respond to questions about it. All other useapi.net services remain fully available.

Cancel your subscription

- Card (Stripe) — open the Stripe Customer Portal and cancel there.
- Crypto (BoomFi) — open the BoomFi Customer Portal and cancel there.

Request a prorated refund

If you'd like a prorated refund, email support@useapi.net from your account email.
The refund is prorated from the date of your request — it covers the unused time remaining in your current billing period, not the full period.

Thanks for your patience and support.

PixVerse API v2 — Seedance 2.0 Mini and omni audio references:

• New video model seedance-2.0-mini — the most cost-effective Seedance 2.0 (480p/720p, 4–15s), available for text-to-video, image-to-video, omni fusion, and transition.
• Seedance omni (fusion) now accepts up to 3 reference videos (video_1_path…video_3_path, referenced as @video1…@video3) and 3 reference audios (audio_1_path…audio_3_path, referenced as @audio1…@audio3) across the whole Seedance 2.0 family — drive motion from a reference clip and speech or sound from a reference audio onto a real human face. Each reference runs 2–15s, with the combined video durations and combined audio durations each capped at 15s.

Examples

MiniMax API v1 — Seedance 2.0 Mini, 4K, and expanded omni references on POST videos/create:

• New video model Seedance-2.0-Mini — the cheapest Seedance 2.0 tier (480p/720p, 4–15s), for text-to-video, image-to-video, and omni references.
• Seedance-2.0 adds a 4K resolution tier alongside 480p/720p/1080p. Seedance-2.0-Fast and Seedance-2.0-Mini stay at 480p/720p.
• Auto is now the default aspectRatio for Seedance (matches the input), alongside 21:9, 16:9, 4:3, 1:1, 3:4, 9:16.
• Omni references expanded to 9 images (@Image1…@Image9), 3 videos (@Video1…@Video3), and 3 audios (@Audio1…@Audio3). Each video and audio reference runs 2–15s, with the combined video + audio length capped at 15s. A clear real human face in a video reference is rejected by the content filter — image references may contain real faces.
• Seedance prompts accept up to 7,500 characters.

Examples

Google Flow API v1 — more generous automatic captcha solving:

• The one-time free captcha credit grant on your first Google Flow account is now 300 (was 100), powered by CapSolver. These credits are used automatically when no provider API keys are configured — check the remaining balance via GET /accounts/captcha-providers.
* The default number of captcha retry attempts is now 5 (was 3), used when a request sends no captchaToken, captchaRetry, or captchaOrder.

MiniMax API v1 — connect your account from your own browser:

• New one-page Connect MiniMax flow links your Hailuo AI account with a quick copy-and-paste — no more hunting for the url and token in the browser network tab. Copy the snippet, run it on hailuoai.video while logged in, and it captures your session and adds the account. See Setup MiniMax for the steps.