TokenMix Research Lab · 2026-05-18
Veo 4 Release Date: 70% Odds for I/O 2026, Veo 3.1 Lite Live
Last Updated: 2026-05-18 Author: TokenMix Research Lab Data checked: 2026-05-18
Google has not officially announced Veo 4 yet. As of May 18, 2026, Google's public model pages, Vertex AI documentation, and Gemini API pricing page still list Veo 3.1 as the latest video generation family. The most recent official expansion is Veo 3.1 Lite, published in April 2026.
That said, the release window is unusually tight. Google I/O 2026 starts on May 19, and Google's own save-the-date post says the event will cover "latest AI breakthroughs" across Gemini, Android, and more. Given the Veo release cadence — Veo in May 2024, Veo 2 later in 2024, Veo 3 in 2025, and Veo 3.1 in late 2025 / early 2026 — Veo 4 is the obvious video-model candidate to watch.
Below is what is confirmed, what is still rumor, what Veo 4 would need to beat, and how teams should prepare API integrations before the model appears.
Table of Contents
- Quick Verdict: Veo 4 Launch Snapshot
- What Google Has Actually Confirmed
- Veo 4 Release Date: Why Google I/O Is the Main Window
- Current Veo 3.1 API Pricing: The Baseline Veo 4 Must Beat
- Expected Veo 4 Capabilities vs Veo 3.1
- Cost Per Video: Veo 3.1 Today vs Likely Veo 4 Pricing
- How to Access Veo Models Today
- Migration Checklist: Veo 3.1 → Veo 4
- FAQ
Quick Verdict: Veo 4 Launch Snapshot
| Claim | Status | Source / Confidence |
|---|---|---|
| Veo 4 is officially released | Not confirmed | No official Google / DeepMind / Vertex AI page yet |
| Latest official Veo model | Confirmed: Veo 3.1 | Google DeepMind Veo page |
| Latest lightweight variant | Confirmed: Veo 3.1 Lite | Google DeepMind model card, April 2026 |
| Google I/O 2026 date | Confirmed: May 19-20, 2026 | Google official save-the-date |
| Veo 4 at Google I/O | Prediction | 70% probability based on cadence and timing |
| Current API model IDs | Confirmed for Veo 3.1 | veo-3.1-generate-preview, veo-3.1-fast-generate-preview, veo-3.1-lite-generate-preview |
| Veo 4 API model ID | Unknown | Likely veo-4.0-generate-preview, not confirmed |
| Current Veo 3.1 price | Confirmed | $0.40/sec standard, $0.10/sec fast, $0.05/sec lite at 720p |
| Veo 4 pricing | Unknown | Likely premium at launch if quality jumps |
Bottom line: Veo 4 is not live yet, but the probability of a near-term announcement is high. If Google announces it at I/O, expect consumer access through Gemini / Flow first, followed by staged API access through Gemini API and Vertex AI.
What Google Has Actually Confirmed
Google's official Veo page currently positions Veo 3.1 as its state-of-the-art video generation model. It highlights text-to-video, image-to-video, native audio, realistic physics, first-and-last-frame generation, object insertion, and production workflows through Flow.
The Gemini API documentation describes Veo 3.1 as capable of generating high-fidelity 8-second 720p or 1080p videos with natively generated audio. It also lists newer workflow controls:
| Capability | Veo 3.1 Status |
|---|---|
| Text to video | Supported |
| Image to video | Supported |
| Native audio | Supported |
| First + last frame generation | Supported |
| Video extension | Supported in preview routes |
| Reference images | Supported in Gemini API docs |
| 4K output | Priced on Gemini API |
| Long-form generation | Still limited; workflows require stitching / extension |
The key point: there is no official Veo 4 model card, no Vertex AI Veo 4 model ID, and no Gemini API pricing entry for Veo 4 as of this update.
Veo 4 Release Date: Why Google I/O Is the Main Window
Google I/O 2026 runs May 19-20. Google says the event will cover AI breakthroughs and updates across Gemini and other products. That makes I/O the highest-probability launch window for a major video model.
The historical cadence supports the theory:
| Model | Public Timing | Notes |
|---|---|---|
| Veo | May 2024 | First major public Veo announcement |
| Veo 2 | Late 2024 | Quality and control improvements |
| Veo 3 | 2025 | Native audio became a headline capability |
| Veo 3.1 | Late 2025 / early 2026 | Stronger control, API expansion, pricing tiers |
| Veo 3.1 Lite | April 2026 | Lower-cost video generation variant |
| Veo 4 | Not announced | Most likely candidate for next flagship jump |
Our probability estimate:
| Release Window | Probability | Reason |
|---|---|---|
| Google I/O 2026 keynote | 70% | Best venue for Gemini / Flow / Veo announcement |
| June-July 2026 | 20% | Allows staged post-I/O product rollout |
| Later 2026 | 10% | Possible if safety, licensing, or compute delays remain |
This is a prediction, not a confirmed date.
Current Veo 3.1 API Pricing: The Baseline Veo 4 Must Beat
Google's Gemini API pricing page lists Veo 3.1 on the paid tier with per-second billing.
| Model | 720p | 1080p | 4K |
|---|---|---|---|
| Veo 3.1 Standard | $0.40/sec | $0.40/sec | $0.60/sec |
| Veo 3.1 Fast | $0.10/sec | $0.12/sec | $0.30/sec |
| Veo 3.1 Lite | $0.05/sec | $0.08/sec | Not supported |
| Veo 3 | $0.40/sec | Not separately listed | Fast supports 4K pricing |
| Veo 2 | $0.35/sec | Not separately listed | Not listed |
For an 8-second clip:
| Model | 8s 720p Cost | 8s 1080p Cost | 8s 4K Cost |
|---|---|---|---|
| Veo 3.1 Standard | $3.20 | $3.20 | $4.80 |
| Veo 3.1 Fast | $0.80 | $0.96 | $2.40 |
| Veo 3.1 Lite | $0.40 | $0.64 | N/A |
If Veo 4 launches as a flagship model, expect it to either keep the $0.40/sec standard price to accelerate adoption, or launch at a premium tier if the quality jump is obvious.
Expected Veo 4 Capabilities vs Veo 3.1
Google has not published Veo 4 capabilities. But based on the gaps in Veo 3.1 and the direction of the video-generation market, the likely upgrade areas are clear.
| Area | Veo 3.1 Today | What Veo 4 Needs |
|---|---|---|
| Clip length | Mostly short clips, commonly 4-8s | Longer coherent shots, fewer extension artifacts |
| Audio | Native audio supported | Better dialogue timing, cleaner ambient sound |
| Character consistency | Improved, still workflow-dependent | Reliable multi-shot identity retention |
| Scene control | First/last frame, reference images, object insertion | More granular camera, motion, and edit controls |
| Physics | Strong internal benchmark claims | Fewer object-morphing and continuity errors |
| Editing | Flow workflows help | True inpainting / selective re-rendering |
| API usability | Preview model IDs and per-second billing | Stable model IDs, clearer quotas, better batch economics |
The biggest practical upgrade would not be "prettier video." It would be controllability. Production teams need to revise one object, keep a face consistent across shots, or extend a scene without restarting from scratch. That is where Veo 4 can create the biggest workflow jump.
Cost Per Video: Veo 3.1 Today vs Likely Veo 4 Pricing
Video generation cost scales differently from text models. A single "small" generation can cost more than thousands of LLM calls.
| Workload | Veo 3.1 Lite | Veo 3.1 Fast | Veo 3.1 Standard |
|---|---|---|---|
| 100 × 8s 720p clips | $40 | $80 | $320 |
| 1,000 × 8s 720p clips | $400 | $800 | $3,200 |
| 10,000 × 8s 720p clips | $4,000 | $8,000 | $32,000 |
For production workloads, the key question is not only model quality. It is rejection rate.
If Veo 4 costs 25% more but cuts failed generations by 50%, it can be cheaper per usable clip. If it costs 2x more and still needs the same number of retries, Veo 3.1 Fast or Lite remains the better bulk option.
How to Access Veo Models Today
There are three practical routes.
1. Gemini App / Flow
Best for creators and non-technical users. Flow is Google's AI filmmaking interface built around Veo, Imagen, and Gemini workflows.
Use this route if you need:
- Prompt-to-video experimentation
- Storyboarding
- Visual iteration
- Creative review before API automation
2. Gemini API
Best for developer workflows that need programmatic video generation. Google currently lists Veo 3.1 preview models on Gemini API pricing:
veo-3.1-generate-preview
veo-3.1-fast-generate-preview
veo-3.1-lite-generate-preview
3. Vertex AI
Best for enterprise Google Cloud teams that need project-level quotas, IAM, audit logs, and production governance. Vertex AI documentation lists Veo models including:
veo-2.0-generate-001
veo-3.0-generate-001
veo-3.0-fast-generate-001
veo-3.1-generate-preview
veo-3.1-fast-generate-preview
Via TokenMix.ai
TokenMix.ai tracks model availability, pricing, and API routing across major AI providers. When Veo 4 becomes available through supported upstream API channels, the cleanest integration pattern is to avoid hardcoding provider-specific model IDs in application logic.
Use a config-driven model switch:
VIDEO_MODEL = "veo-3.1-fast-generate-preview"
Then promote to Veo 4 only after evals confirm better cost per accepted clip.
Migration Checklist: Veo 3.1 → Veo 4
| Action | When | Effort |
|---|---|---|
| Save current Veo 3.1 prompts and outputs | Now | 1 hour |
| Build a 50-prompt video eval set | Now | 2-4 hours |
| Track accepted vs rejected generations | Now | Half day |
| Separate creative prompts from API parameters | Now | 1 day |
| Add model ID as config, not code | Now | 30 minutes |
| Compare Veo 4 against Veo 3.1 Fast / Standard | Launch day | 1 day |
| Measure cost per usable clip, not cost per generation | Launch week | 1-2 days |
| Keep Veo 3.1 Lite for bulk drafts | After launch | Ongoing |
Do not migrate purely because Veo 4 is newer. Migrate if it reduces retries, improves controllability, or unlocks a workflow Veo 3.1 cannot handle.
FAQ
Is Veo 4 released?
No. As of May 18, 2026, Google has not officially announced Veo 4. The latest official family is Veo 3.1, with Veo 3.1 Lite published in April 2026.
When is the Veo 4 release date?
The most likely window is Google I/O 2026 on May 19-20, but that is a prediction. Google has confirmed the event dates, not a Veo 4 launch.
What is the latest official Veo model?
Veo 3.1 is the latest official flagship video generation model listed by Google. Veo 3.1 Lite is the latest lower-cost variant.
What will Veo 4 improve?
The most likely improvements are longer coherent clips, better character consistency, stronger audio-video synchronization, fewer physics artifacts, and more precise editing controls. None of these are confirmed for Veo 4 yet.
How much will Veo 4 cost?
Unknown. Veo 3.1 currently costs $0.40/sec for standard 720p or 1080p output, $0.10-$0.12/sec for Fast, and $0.05-$0.08/sec for Lite. Veo 4 may launch at the same standard price or at a premium.
Can I use Veo 4 through the API?
Not yet. There is no public Veo 4 API model ID. Current API routes use Veo 3.1 preview model IDs.
Should I wait for Veo 4 or use Veo 3.1 now?
Use Veo 3.1 now if you need production video generation today. Prepare your prompts and eval suite so you can test Veo 4 immediately if it launches.
Related Articles
- Veo 4 in 2026: It's Not Released, So What Are You Buying?
- GPT-5.5 (Spud) Released: $5/$30 API Pricing & Benchmarks 2026
- Claude API Pricing 2026: Opus, Sonnet, Haiku Costs Compared
- Doubao API Setup 2026: 19 Models, $0.022/M Floor, Python Guide
- Gemini API Pricing 2026: Text, Image, Audio, Video Models Compared