← All models

Veo 3

Video Retired Full comparison ↗

google-deepmind/veo-3 · by Google DeepMind · Diffusion transformer (DiT)

Pricing — 1 offering(s)

Google DeepMind (Gemini API)

About this offering: Retired: Google shut down Veo 3 (veo-3.0-generate-001 and veo-3.0-fast-generate-001) on 2026-06-30, so the prices shown are its last published rates. It generated video with native audio, at Standard and Fast tiers. Google pointed users to Veo 3.1 (preview) or the generally available video models on its Gemini Enterprise Agent Platform.

Standard

Retired — no longer sold · last price $0.40 / second, until 2026-06-29

  • $0.40 / second Historical 2026-06-13 → 2026-06-29

Fast 720p

Retired — no longer sold · last price $0.10 / second, until 2026-06-29

  • $0.10 / second Historical 2026-06-13 → 2026-06-29

Fast 1080p

Retired — no longer sold · last price $0.12 / second, until 2026-06-29

  • $0.12 / second Historical 2026-06-25 → 2026-06-29

Fast 4K

Retired — no longer sold · last price $0.30 / second, until 2026-06-29

  • $0.30 / second Historical 2026-06-13 → 2026-06-29

Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.

See how Veo 3 fits into a cost-aware routing setup

See how →

Capability profile

How Veo 3 rates across core capability dimensions, with the task-level evidence behind each rating.

Prompt adherence Strong

Strong text-video alignment; handles complex compositional prompts.

temporal consistency Strong

Subjects and backgrounds are stable across the full clip duration.

motion quality Strong

Industry-leading as of release; natural human motion and physics.

native audio Strong

First major commercial T2V model with native synchronized audio generation.

camera control Moderate

Responds to cinematic camera terminology in prompts; no explicit control API.

Resolution ceiling Strong

4K available via the Fast 4K tier ($0.30/s).

Inference speed Moderate

Fast tier significantly reduces latency at lower quality.

clip duration ceiling Moderate

Max 8 seconds per call.

cost efficiency Weak
video editing Weak

T2V and I2V only via Gemini API; no editing of existing clips exposed at the model API level.

Ratings are estimated — limited independent data is available for this model.

Benchmarks

Benchmark Score Config Source
vbench 87.2 Approximate; verify against current leaderboard as scores may be updated. leaderboard ↗

Operator guidance

Retired: Google shut down Veo 3 on 2026-06-30. For native-audio video generation, look at Veo 3.1 (preview) or another current model with built-in audio.

Use cases

Limitations

Citations