Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.apiyi.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Key Highlights

  • Google AI Studio official-relay channel: Transparent passthrough to Google Veo 3.1 async endpoints; model IDs match upstream exactly
  • Two models: veo-3.1-fast-generate-preview ($0.3/req) and veo-3.1-generate-preview ($1.2/req) — per-request billing, independent of duration / resolution
  • Full capabilities: flexible 4 / 6 / 8 second durations + 720p / 1080p / 4k resolution tiers + landscape/portrait + native synchronized audio
  • Zero onboarding friction: Default group works, Pay-per-request or Pay-as-you-go Priority Tokens both accepted (pure Pay-as-you-go not supported), no dedicated group switch — existing Tokens drop in with just a base_url change
  • Failures are free: only status=completed tasks are billed; failed / canceled / content-policy rejected are all free
  • Complementary to the existing Reverse channel: VEO 3.1 (Reverse) stays online at $0.15 starting, with streaming sync and frame-to-video; the new Official channel targets official-grade quality and stability. See the comparison matrix

​
Background

Google released the Veo 3.1 series in 2026 as a flagship cinematic video generation engine, featuring 4K output, native synchronized audio, complex camera motion, and exceptional temporal consistency. The upstream offers two variants:
  • veo-3.1-generate-preview — highest-quality standard tier
  • veo-3.1-fast-generate-preview — cost-optimized fast tier
APIYI previously offered Veo 3.1 via the Reverse channel (VEO 3.1) — Google Flow integration starting at $0.15, with streaming sync and frame-to-video, ideal for cost-sensitive workloads. But the Reverse channel, being reverse-engineered, differs from upstream on quality stability, model ID consistency, and enterprise compliance. To close this gap, the APIYI team launches the VEO 3.1 Official Relay channel: a transparent passthrough to Google AI Studio, with model IDs, response fields, and constraints matching Google upstream exactly. Available 2026-05-21 (UTC+8).

​
Detailed Walkthrough

​
Key Features

🎬 Native Synchronized Audio

Veo 3.1 natively outputs video with synchronized audio (ambient, dialogue, score) — no separate audio post-production. Describe audio intent in the prompt, do not pass generateAudio.

📐 Three Resolution Tiers

720p / 1080p / 4k — uniform per-request price. Landscape 16:9 and portrait 9:16 toggle freely.

⏱️ Flexible 4 / 6 / 8 Seconds

duration string enum; per-request billing means duration does not affect price. 1080p / 4k tiers require 8 seconds.

🎯 Precise Instruction Following

Veo 3.1 leads its tier on camera motion, object physics, and character expression fidelity — rich camera-language keyword support.

​
Key Selling Point: Drop-In Onboarding

Versus Sora 2 Official Relay (which requires the dedicated Sora2Official group + Pay-as-you-go Priority Token only), VEO 3.1 Official runs on the Default group and accepts both Pay-per-request and Pay-as-you-go Priority Tokens — existing Tokens drop in without any config changes: 
{/* Single base_url change; no group switch, no specialized Token needed */}
from openai import OpenAI
client = OpenAI(api_key="sk-your-api-key", base_url="https://api.apiyi.com/v1")

resp = client.post(
    "/videos",
    body={
        "model": "veo-3.1-fast-generate-preview",
        "prompt": "A coastal lighthouse at dusk, slow push-in, ocean ambience, cinematic lighting",
        "duration": "8",
        "size": "1280x720",
        "metadata": {"resolution": "720p", "aspectRatio": "16:9"}
    },
    cast_to=dict
)

​
Models and Pricing — 60%+ Cheaper than Google Official

APIYI uses pay-per-request billing — flat price within supported duration/resolution combos, no surcharge for longer or higher-res output. Per ai.google.dev/gemini-api/docs/pricing public rates, Google’s official Veo 3.1 charges per second; the discounts below are computed for 8-second videos.
ModelAPIYI PriceGoogle Official 8s 1080pGoogle Official 8s 4K
veo-3.1-fast-generate-preview$0.3 / req$0.96
68.8% off		$2.40
87.5% off
veo-3.1-generate-preview$1.2 / req$3.20
62.5% off		$4.80
75.0% off
Per-request billing by model name; duration (4/6/8 sec), resolution (720p/1080p/4k), and whether input_reference is provided do not affect price — picking 4K costs the same as 720p. Only status=completed tasks are charged, and top-up bonuses further reduce effective cost.

​
Technical Specs

Dimensionveo-3.1-fast-generate-previewveo-3.1-generate-preview
Supported duration"4" / "6" / "8" sec (string)Same
Supported resolution720p / 1080p / 4kSame
Supported aspect16:9 / 9:16Same
EndpointPOST /v1/videos (async only)Same
Audio✅ Synchronized audio✅
Image-to-video✅ (1 input_reference image)✅
Typical render time720p 60–90s · 1080p 80–120s · 4K 5–6 minSame

​
Three-Step Async Flow

Step 1: POST /v1/videos          → returns task_id + status="queued"
Step 2: GET /v1/videos/{task_id} → poll every 8 sec, until status="completed"
Step 3: GET /v1/videos/{task_id}/content → download MP4 binary

​
Real-World Applications

​
When to Use Which Channel

✅ Use Official

  • Final client ad delivery
  • 4K cinematic clips
  • Quality stability / physics consistency priority
  • Overseas teams, existing Token unchanged
  • Strong instruction following (camera motion, expression)

🔄 Use Reverse (VEO 3.1)

  • Short-video matrix batch production ($0.15 unit price)
  • Frontend live display needing streaming sync progress
  • Need frame-to-video (Frame-to-Video) capability
  • Extreme cost-sensitivity
See the Official vs Reverse decision tree for full scenario-by-scenario recommendations.

​
Three Most Common Pitfalls

  1. duration must be a string "4" / "6" / "8"; passing a number fails with parse_request_failed
  2. Do not pass generateAudio — upstream returns INVALID_ARGUMENT. Audio intent goes in the prompt
  3. At 1080p / 4k, duration must be "8" — "4" / "6" will be rejected upstream
Full code samples and 20+ FAQs in the Official Overview and Text-to-Video Playground.

​
Pricing and Availability

  • Launch time: 2026-05-21 (UTC+8)
  • Group: Default (1x, no switching needed)
  • Billing model: Pay-per-request ✅ / Pay-as-you-go Priority ✅ (pure Pay-as-you-go ❌ not supported)
  • Top-up bonuses: Stack with Top-Up Promotions for further savings
  • Available gateways: api.apiyi.com (primary) + vip.apiyi.com / b.apiyi.com (backup)

​
Summary and Recommendations

VEO 3.1 Official is APIYI’s official-relay channel for Veo 3.1 — a transparent passthrough to Google AI Studio, positioned similarly to Sora 2 Official but with lower onboarding friction:
  • ✅ Model IDs and response fields match Google upstream exactly
  • ✅ Default group + Pay-per-request or Pay-as-you-go Priority Tokens work (pure Pay-as-you-go not supported) — existing Tokens require no config change
  • ✅ Flexible 4 / 6 / 8 seconds + 720p / 1080p / 4k resolution tiers
  • ✅ Failed tasks are not billed; charged only on successful completion
  • ✅ Coexists with the existing Reverse channel — complementary by scenario
Recommend starting with veo-3.1-fast-generate-preview ($0.3 for 4–8 sec) to validate your prompts, then switching to veo-3.1-generate-preview for final delivery once the look is locked.
Google official model page: ai.google.dev/gemini-api/docs/models/veo-3.1-generate-preview · Google video generation hub: ai.google.dev/gemini-api/docs/video