Asset Generator is now live on Teno — one chat-based studio that hosts our full curated stack of image and video models, with opinionated defaults for each ad-creative job. The picker exposes 7 image models (defaulting to Nano Banana Pro) and 7 video models (defaulting to Seedance 2.0); the rest are one click away when the job calls for them. One workspace, one credit pool, one thread per creative idea — switch between image and video without leaving the conversation.
Key Takeaways
- Curated defaults across a 14-model stack. Asset Generator exposes 7 image and 7 video models in one picker, with opinionated defaults per job — Nano Banana Pro for image, Seedance 2.0 for video — so most teams ship without running N×N model comparisons.
- Chat-thread iteration is the primitive. Each Asset Generator chat preserves prompts, references, and outputs across prompts and model swaps — not a "compare 30 models side-by-side" grid.
- One credit pool across all 14 models. As of May 2026, image generations range from 1 to 3 credits and video clips from 10 to 26 credits per 5 seconds, depending on tier and model; see pricing for live conversion.
- Hub for the May 5 model launches. Asset Generator is where GPT Image 2 and Seedance 2.0 actually live together; this post is the studio-level view.
This post is the canonical Teno write-up on Asset Generator: what it is, which model fits which job, how it compares to Creatify (the closest multi-model alternative), and what it costs. For the wider category framework Asset Generator sits inside, see our AI ad creative platform guide.
What is an AI asset generator, and how is Asset Generator on Teno different?
An AI asset generator is a single workspace where teams generate ad images and short videos across multiple AI models without switching between tools. Asset Generator on Teno is the curated, ad-creative-focused version of that category: 7 image and 7 video models in one chat-based studio, with opinionated defaults per job so most briefs never need a side-by-side model comparison to ship.
The two design choices that separate it from the broader category:
- Curated defaults, not just a long picker. Most multi-model AI generators advertise model count as the headline (20+, 30+, sometimes 50+) and leave you to pick. Asset Generator's curation runs at the default level — Nano Banana Pro for image and Seedance 2.0 for video are the picks for the vast majority of ad-creative jobs out of the box. The full 14-model picker (Flux 2 Pro / Max, Seedream 4.5 / 5 Lite, Nano Banana 2, Sora 2 Pro, and the Fast tiers of Veo and Seedance) is there for the cases where you need a typography model, a long-shot cinematic model, or a fast-tier draft — without leaving the chat.
- Chat threads, not comparison grids. Most competitor UIs are built around "generate the same prompt across N models, compare side-by-side, pick a winner." That maps to a one-shot exploration job. Real ad-creative work is iterative: a brand thread runs for hours, model choice changes mid-thread, references accumulate. Threads model that workflow; grids do not.
For the broader 4-layer framework Asset Generator sits inside as Layer 1, see the AI ad creative platform guide. For the model-by-model deep dives, the two launch posts below are the canonical references.
Which AI models are inside Asset Generator, and when should you use each?
The Asset Generator 2026 catalog ships with 14 models as of May 22, 2026: 7 image models and 7 video models. The five names you'll reach for most often — Nano Banana Pro, GPT Image 2, Seedance 2.0, Veo 3.1, and Kling O3 Pro — get the spotlight below, and the 9 alternates fill out the picker for narrower jobs (Flux 2 Pro / Max for editorial stills, Seedream 4.5 / 5 Lite for ByteDance-family image consistency, Nano Banana 2 for high-volume backgrounds, Sora 2 Pro for photoreal hero shots, plus the Fast tiers of Veo and Seedance for draft mode).
Image models
The defaults plus the high-recall picks ad teams reach for most:
- Nano Banana Pro (default) — current default for image generations. Commercial-grade character and product consistency across a campaign, which makes it the right pick when the same product or character has to recur across many frames.
- GPT Image 2 — the typography and precise-edits model. Renders accurate text across 50+ scripts including Arabic, Japanese, Korean, and Devanagari (Microsoft, Azure AI Foundry Blog, April 2026). It is also the model to reach for on multi-reference edits where spatial constraints have to hold. See the GPT Image 2 launch post for the full capability map.
Plus 5 alternates in the picker for narrower jobs:
- Nano Banana 2 — Google's earlier-generation Nano Banana at 2 credits per image; useful for high-volume background and lifestyle work that doesn't need Pro-grade consistency.
- Flux 2 Pro — Black Forest Labs' photoreal generalist; strong for editorial-style stills with painterly lighting.
- Flux 2 Max — Flux's premium tier for the cases where Pro doesn't quite hold; reach for it when the brand requires Flux's specific aesthetic at maximum quality.
- Seedream 4.5 — ByteDance's image model in the same family as Seedance; convenient when the still has to feed a Seedance image-to-video downstream with consistent lighting.
- Seedream 5 Lite — Seedream's fast-and-cheap tier; useful for rapid layout exploration before committing credits to a Pro-tier render.
Video models
The three picks that cover the majority of ad-team video jobs:
- Seedance 2.0 (default) — current default for video generations. ByteDance's Seedance 2.0 supports native audio-video generation and start+end-frame image-to-video, with 4–15 second output (ByteDance Seed, 2026). For UGC, product reveals, and weekly variant testing, it is the strongest single pick. The Seedance 2.0 launch post covers pricing, and the Seedance 2.0 ad-creative guide covers prompt patterns.
- Veo 3.1 — talking-head and lip-sync precision. When a single tight shot needs the mouth to match the audio frame-by-frame, Veo still holds a narrow edge.
- Kling O3 Pro — long, stylized cinematic motion. The right pick for slow pulls, establishing shots, and high-style hero clips where the camera move is the point.
Plus 4 alternates in the picker:
- Seedance 2.0 Fast — the Seedance 2.0 family's faster-and-cheaper tier (16 credits / 5s vs 20 on standard). Use for rapid concept testing before promoting the winner to standard.
- Veo 3.1 Fast — Veo's fast-and-cheaper tier (12 credits / 5s); use when Veo's lip-sync edge matters but you're still in draft mode.
- Kling O3 Standard — Kling O3 at 10 credits / 5s; supports 15s clips and native audio. The right pick when you want Kling's cinematic motion at a lower cost than Pro.
- Sora 2 Pro — OpenAI's video model. Strong on photoreal stills-into-motion hero shots; the right pick when the seed image is polished and the prompt is camera moves only.
The decision matrix
The five picks below cover what most ad-creative jobs end up using — the table to bookmark when reaching into an Asset Generator chat. The 9-row alternates table further down covers the narrower situations.
| Model | Modality | Best for | Credits (representative) | When to skip |
|---|---|---|---|---|
| Nano Banana Pro | Image | Recurring product / character consistency | 3 / image | When typography or precise multi-reference edits are the job |
| GPT Image 2 | Image | Typography, multilingual social cards, multi-reference edits | 3 / image | When you need maximum character likeness across a campaign |
| Seedance 2.0 | Video | Native audio-video; start+end-frame image-to-video | 20 / 5s 720p clip with audio | When tight lip-sync is the headline deliverable |
| Veo 3.1 | Video | Talking-head precision; clean prompt-to-motion in short shots | 26 / 5s 720p clip with audio | When credit cost matters more than lip-sync precision |
| Kling O3 Pro | Video | Long, stylized cinematic motion; establishing shots | 12 / 5s 720p clip | When you need native audio in the same generation |
More in the picker — 9 alternates for narrower jobs:
| Model | Modality | Best for | Credits (representative) |
|---|---|---|---|
| Nano Banana 2 | Image | High-volume background / lifestyle creative | 2 / image |
| Flux 2 Pro | Image | Editorial-style stills with painterly lighting | 1 / image |
| Flux 2 Max | Image | Brand-specific Flux aesthetic at premium quality | 1 / image |
| Seedream 4.5 | Image | Image that feeds Seedance I2V with consistent lighting | 1 / image |
| Seedream 5 Lite | Image | Rapid layout exploration before a Pro-tier render | 1 / image |
| Seedance 2.0 Fast | Video | Quick Seedance draft before promoting to standard | 16 / 5s 720p clip with audio |
| Veo 3.1 Fast | Video | Veo's lip-sync edge in draft mode | 12 / 5s 720p clip |
| Kling O3 Standard | Video | Kling cinematic motion + 15s + native audio at lower cost than Pro | 10 / 5s clip |
| Sora 2 Pro | Video | Photoreal stills-into-motion hero shots | 20 / 5s clip |
Numbers above are from the Teno model catalog as of May 2026. Video credit costs scale with duration and resolution; the figures are the base rate at the configurations most ad teams use.
For a visual on cost spread among the three most-used video picks — meaningful because they vary by more than 2× — see below. The full 7-video catalog spans 10 credits (Kling O3 Standard) to 26 credits (Veo 3.1) per 5-second clip.
How does the chat-based workflow change ad iteration?
Asset Generator organizes work into chat threads — one thread per creative concept. Inside the thread, prompts, attached references, and prior outputs all stay in context, so switching models mid-thread does not reset the conversation. That is a deliberate choice: most multi-model AI generators are built around a comparison grid ("generate prompt P across models A, B, C, pick a winner"), which optimizes for one-shot exploration. Real paid-creative work is iterative, not one-shot.
When chat beats side-by-side comparison
The grid wins when you genuinely don't know which model fits the job and need a quick bake-off across 4–6 candidates on one prompt. We see that maybe 10% of the time. The other 90% is some version of: "this concept is the one — render 5 variations, swap the background, regenerate the hero, then promote the winner to video." That work is iterative. It needs the brand context, the references, the prior outputs, and the prompt history all to stay attached to one workspace.
Chat threads also reflect how ad teams talk about creative internally. "The new Nordic skincare hero, draft 4" maps cleanly to a single thread named "Nordic skincare hero." A side-by-side grid doesn't have a noun for that unit of work; it has a screen of swatches that someone has to manually curate into a folder.
The trade-off is straightforward. If your job is single-shot model evaluation — "which of these six models renders the best version of one prompt?" — Asset Generator's chat-thread UI is heavier than you need. That use case is real but rare in production paid-creative workflows.
When do you actually need both image and video in one studio?
Three repeatable ad-creative jobs collapse from two-tool workflows to one Asset Generator thread once image and video share a workspace. Each one is a job where the still and the motion clip have to share brand context, reference frames, or both — and where bouncing between two tools loses that context.
The three jobs:
- Hero image → image-to-video transition. Generate a polished hero still in Nano Banana Pro or GPT Image 2, then feed it (and an optional end-frame) into Seedance 2.0's start+end-frame image-to-video. Same thread, one credit pool, no asset hand-off.
- Campaign set from one brief. A single prompt produces a hero image plus 3 short video variants for testing — for example, one Seedance 2.0 founder-cam variant, one Veo 3.1 talking-head, one Kling O3 Pro cinematic pull-back. All three keep the same brand context.
- Image-first composition lock, then video promotion. Iterate the layout, lighting, and copy as a still until the concept feels right, then promote the winning frame to motion. Cheaper than running video generations from prompt zero, and easier for stakeholders to review.
The version of this workflow that costs you the most is the two-tool version: image tool A, video tool B, manual file shuttling, lost references, no shared prompt history. The version that scales is one thread.
Asset Generator vs Creatify: which fits which team?
Both Asset Generator and Creatify's Asset Generator are multi-model AI generators for ad creatives, but they make opposite trade-offs on two design choices: how big the model catalog should be, and what the iteration UI primitive is. Both are defensible choices for different teams.
| Dimension | Asset Generator on Teno | Creatify Asset Generator |
|---|---|---|
| Model catalog | 14 (7 image + 7 video) with explicit defaults per job | 30+ image and video models per their product page (May 2026); no published defaults |
| Iteration primitive | Chat threads with persistent context | Side-by-side comparison grid |
| Default image model | Nano Banana Pro | Their picker; defaults vary |
| Default video model | Seedance 2.0 | Their picker; defaults vary |
| Free-tier shape | Monthly credits, no card; both modalities included | Free plan with monthly credits per their product page; export watermark applies |
| Best fit team | Teams running ongoing iteration on a defined creative idea | Teams doing exploratory bake-offs across many models |
Where Creatify genuinely wins: their larger catalog exposes more experimental and niche models — useful if your workflow is "test the same prompt across 8 candidates and pick a winner," or if your brand depends on a specific niche model. The catalog gap is real (14 vs 30+), and we'd point you to Creatify if model coverage breadth is the headline requirement.
Where Teno's trade-off bites: a 14-model catalog cannot cover every experimental or niche model the moment it ships. Every model in our picker is one we'd actively recommend for an ad-creative job — so SDXL variants, NSFW-permissive picks, and fresh research-lab releases tend to land in Creatify before they land here. Our bet is that opinionated defaults plus chat-thread iteration is what teams want once they leave the exploration phase and start producing for an active campaign. If your team is mostly in the exploration phase and bigger model coverage is the priority, Creatify is the better tool for the job today.
The honest one-liner: Asset Generator is for teams iterating on a known creative idea. Creatify Asset Generator is for teams comparing across many models. Neither pitch is wrong; they are sized for different jobs.
How does Asset Generator compare to Teno's Product Photo and Product Video?
Product Photo and Product Video are purpose-built rails for one specific shot type each; Asset Generator is the flexible studio for everything else. The decision rule is simple — if your asset has a single product photo at the center and your job is a known shot type, start on the rail. If you're exploring, need typography, or want multi-reference work, start in Asset Generator.
| Surface | What it's built for | Model choice | When to use it |
|---|---|---|---|
| Product Photo | Studio / lifestyle / flat-lay / luxury shots from one product image | Fixed pipeline | One product, a known shot preset, fast iteration |
| Product Video | Orbit / turntable / glide / close-up motion from one product image | Fixed pipeline | One product, a known motion preset, fast iteration |
| Asset Generator | Open-ended image and video creative; multi-reference; multilingual; mixed | Full 14-model picker | Exploration, typography, multi-reference, mixed-mode |
The three surfaces share one credit pool and one account. Picking the right one is mostly about whether your job fits a known shot preset (use Product Photo / Product Video) or needs the open studio (use Asset Generator). Many teams use both: Product Photo for the high-volume catalog work, Asset Generator for the hero campaign creative.
What does it cost to generate an asset on Teno?
Asset Generator is included in Teno's free tier with monthly credits and no credit card required. All 14 models share that credit pool, so the free tier lets you test image and video without per-model subscriptions. Credit cost per generation depends on the model and the configuration — for the full Asset Generator catalog, May 2026 numbers from the Teno model catalog (defaults bolded):
| Model | Modality | Representative job | Credits |
|---|---|---|---|
| Nano Banana Pro (default) | Image | 1 image at default settings | 3 |
| GPT Image 2 | Image | 1 image at high quality | 3 |
| Nano Banana 2 | Image | 1 image at default settings | 2 |
| Flux 2 Pro | Image | 1 image at default settings | 1 |
| Flux 2 Max | Image | 1 image at default settings | 1 |
| Seedream 4.5 | Image | 1 image at default settings | 1 |
| Seedream 5 Lite | Image | 1 image at default settings | 1 |
| Seedance 2.0 (default) | Video | 1 × 5-second 720p clip with native audio | 20 |
| Seedance 2.0 Fast | Video | 1 × 5-second 720p clip with native audio | 16 |
| Veo 3.1 | Video | 1 × 5-second 720p clip with audio | 26 |
| Veo 3.1 Fast | Video | 1 × 5-second 720p clip | 12 |
| Kling O3 Pro | Video | 1 × 5-second 720p clip | 12 |
| Kling O3 Standard | Video | 1 × 5-second 720p clip with native audio | 10 |
| Sora 2 Pro | Video | 1 × 5-second 720p clip | 20 |
A practical read of the table: stills range from 1–3 credits depending on model, and video clips range from 10–26 credits per 5 seconds depending on tier. The cheapest video clip is at minimum ~3× the cheapest image, and a Veo 3.1 clip is 8–26× a Flux 2 still — which is the strongest argument for the "image-first, then promote to video" pattern in §4: lock the composition cheaply, then spend video credits only on the winners. For higher monthly allocations, premium tiers like Seedance 2.0 standard, and live credit-to-USD conversion, see pricing.
Our finding: in the first two weeks of Asset Generator usage on Teno, ad-team workflows averaged about 6 image generations per 1 video generation — which lines up with the credit gap above and with what we expect when "image-first" becomes the default iteration pattern.
Workflow walkthrough — from a brief to a publish-ready ad set
Asset Generator's three-step shape mirrors the marketing page: describe what you want, pick a model and tune the output, iterate and export. The real value lives in step 3, where the chat thread starts paying off. Here is the sequence in practice for a skincare hero brief.
Step 1 — Describe the creative. Open Asset Generator, start a new chat, and write the brief. For our skincare example: "Premium skincare bottle on polished marble, soft golden-hour light with subtle reflection, magazine cover composition." Attach one or two reference shots if the brand has a strong existing look — the composer's chip bar will already show the model and settings you're about to fire.
Step 2 — Pick a model, tune the output. Open the model picker and confirm Nano Banana Pro (the default) — the picker shows it sitting above GPT Image 2, Seedance 2.0, and Kling O3 Pro as the four cards you'll reach for most. Set aspect ratio (1:1 for a square hero) and output count (×3 covers the variation spread without burning credits). Generate 3 variations at 3 credits each — total 9 credits for the first round. An ad-grade product image lands in about 20 seconds, matching the time-to-first-output benchmark from our AI ad creative platform guide.
Step 3 — Iterate and export. Pick the winning hero from the variation grid, then in the same thread, switch the model to Seedance 2.0 and run image-to-video on the winner with an optional end-frame. The video draft typically takes 2–5 minutes per variant. Download the still and the clip; the entire ad set lands inside one chat thread under one campaign name.
Our finding: for the skincare brief above, total credit spend for one hero image (3 variations) plus one 5-second 720p Seedance 2.0 motion variant is 9 + 20 = 29 credits — comfortably inside the free-tier monthly allocation for a first-time concept test.
Frequently asked questions
What is an AI asset generator?
An AI asset generator is a single workspace where teams generate ad images and short videos across multiple AI models without switching tools. Asset Generator on Teno hosts 7 image and 7 video models — including Nano Banana Pro, GPT Image 2, Seedance 2.0, Veo 3.1, and Kling O3 Pro — with opinionated defaults per job, all running on one credit pool inside a chat-based studio.
Can AI generate both images and videos in the same workspace?
Yes. Asset Generator on Teno is an AI image and video generator for ads that supports both modalities in one chat thread. Switch between image and video mid-thread; references, prompts, and outputs stay attached so a single concept can produce a hero image and a short video without leaving the same workspace.
Which AI model is best for ad creatives in 2026?
There is no single best model — the right pick depends on the job. As of May 2026, the decision rule we use on Teno is: Nano Banana Pro for recurring character or product consistency, GPT Image 2 for typography and multi-reference edits, Seedance 2.0 for native audio-video, Veo 3.1 for talking-head lip-sync, and Kling O3 Pro for long cinematic motion. A multi-model AI generator lets you pick per job instead of forcing every brief through one tool.
How do I switch between AI models without losing my work?
Asset Generator on Teno organizes generations into chat threads. Each thread keeps your prompts, attached references, and prior outputs in context, so switching models mid-thread — for example from Nano Banana Pro to GPT Image 2, or from an image model to Seedance 2.0 — does not reset the conversation. The thread remains the unit of work; the model is one of the dials you turn inside it.
Is Teno Asset Generator free to try?
Yes. Asset Generator on Teno is included in the free tier with monthly credits and no credit card required. All 14 models share one credit pool, so the free tier covers exploration across both image and video without per-model subscriptions. For higher monthly allocations and premium tiers like Seedance 2.0 standard, see pricing for plan details.
Try Asset Generator
Asset Generator is live today on every Teno plan, with the free tier open and no credit card required. The fastest test: start a Teno account, open Asset Generator, and run the skincare brief above end-to-end — one image generation pass, one image-to-video promotion, total credit spend of 29. For the wider 4-layer framework Asset Generator sits inside, see the AI ad creative platform guide. For per-model detail, the two May 5 launch posts are the canonical references: GPT Image 2 on Teno and Seedance 2.0 launched on Teno.
Sources
- ByteDance Seed, Official launch of Seedance 2.0, retrieved 2026-05-22.
- Microsoft, Introducing OpenAI's GPT-image-2 in Microsoft Foundry, Azure AI Foundry Blog, retrieved 2026-05-22.
- Creatify, Asset Generator product page, retrieved 2026-05-22 (model count claim).
- Teno model catalog, May 2026 (credit calibration for all 14 Asset Generator models: Nano Banana Pro, Nano Banana 2, GPT Image 2, Flux 2 Pro, Flux 2 Max, Seedream 4.5, Seedream 5 Lite, Seedance 2.0, Seedance 2.0 Fast, Veo 3.1, Veo 3.1 Fast, Kling O3 Pro, Kling O3 Standard, Sora 2 Pro). Internal documentation; live credit-to-USD conversion reflected on /pricing.
