Leonardo.Ai vs Elyumfor Veo 3.1
The deciding difference is the bill: Elyum charges Veo 3.1 only on keep, Leonardo.Ai deducts tokens on generate and does not publish a per-clip price.
Veo 3.1 is Google's cinematic video model with native audio and spoken dialogue. It is built for 4 to 8 second brand spots where the picture, room tone, and line of dialogue have to land together, not as a silent plate you score later. On Elyum it covers text to video, image to video, and reference to video. The look is film-like camera, lighting, and motion in short shots. If the job is a talking product insert or a dialogue-led hero clip, this is the model people reach for.
The model is the same. The bill is not. Leonardo.Ai sells Veo 3.1 (plus Fast/Lite variants) on a Fast Token subscription; third-party video always consumes the full token rate, even on Ultimate at $60/mo, and Leonardo.Ai does not publish a per-second dollar price. Elyum lists Veo 3.1 in credits on the Veo 3.1 page: 21 credits per second at 720p, charged only when you keep a take. A failed generation is never charged on Elyum. Leonardo.Ai support says tokens should restore after a failure, but restoration is not necessarily instantaneous, and users have reported needing to email [email protected] with a generation ID.
Leonardo.Ai is the better desk if you need a shared team workspace, canvas editing, or first-party image models with Relaxed (token-free, slower) generation on Premium and Ultimate. Elyum has none of those. Elyum is one studio and one credit balance across 50+ models, plus MCP and a REST API. There is no timeline editor, no stock footage, no avatars, no mobile app, no team seats, and no community feed. If those are the job, stay on Leonardo.Ai.
Veo 3.1 on Leonardo.Ai
Leonardo.Ai lists Veo 3.1 (plus Fast and Lite variants) on its Veo 3 page. The model runs on Fast Token plans and consumes the full token rate even on Ultimate ($60/mo, 60,000 fast tokens plus a 180,000-token rollover bank). Relaxed generation on Premium and Ultimate applies only to first-party models (Lucid Origin, Phoenix, FLUX, Motion). Veo is excluded. Leonardo.Ai does not publish a per-second or per-clip dollar price. eesel AI's Leonardo.Ai pricing breakdown treats one 8-second Veo 3 clip as roughly 300+ basic first-party image generations in tokens.
Tokens deduct when you hit generate. Leonardo.Ai support says a failed job should restore tokens automatically, but restoration is not necessarily instantaneous. An Adobe Community report describes credits gone after a video that never generated, including one ~800-credit loss, with refunds after emailing [email protected] plus generation ID and timestamp. Trustpilot reviewers call the token model confusing once third-party video is in the mix. Paid generations are private, unmarked, and come with full IP ownership. Free-tier outputs (150 tokens/day, no rollover) carry a watermark and are public by default.
Veo 3.1 on Elyum
On Elyum, Veo 3.1 is priced per second (about 20% under comparable platforms, already in the list price). 720p is 21 credits/second (5s = 104 credits = $1.04; 8s = 166 credits = $1.66). 1080p is 29 credits/second (5s = 144 credits = $1.44; 8s = 230 credits = $2.30). 4K is 60 credits/second (5s = 300 credits = $3.00; 8s = 480 credits = $4.80). Clips are 4 to 8 seconds with native audio. Modes are text to video, image to video, and reference to video. Plans are on Elyum pricing: Free (up to 150 credits once (our algorithm may grant them), no card), Plus $9/mo, Pro $29/mo, Ultra $69/mo.
Every render comes back as a free preview. Credits move only when you KEEP a take. Kill the rest and the refund is automatic and full. A failed generation is never charged. Free accounts get one kill a month; Plus, Pro, and Ultra kill freely. Kept files download clean, no watermark, commercial use included. You can drive the same balance from Claude, ChatGPT, or Cursor through Elyum's MCP server or the REST API.
What a Veo 3.1 take costs
| Scenario | Leonardo.Ai | Elyum |
|---|---|---|
| One 5s take you keep | Not published. Full token rate even on Ultimate; one 8s Veo 3 clip ≈ 300+ basic first-party image gens | 720p: 5 × 20.8 = 105 credits ($1.05). 1080p: 5 × 28.8 = 145 credits ($1.45) |
| Ten 5s 720p takes, keep two | Not published. Tokens deduct on each generate; discarded takes are not a listed refund path | 2 × 105 credits = 210 credits ($2.10). Eight kills refund in full |
| A failed generation | Tokens taken on generate. Auto-restore not necessarily instant; else email [email protected] with ID | Never charged. Failed jobs do not touch the credit balance |
| Unlimited / relaxed generation | Relaxed on Premium/Ultimate is first-party only. Veo always consumes tokens | No unlimited Veo. Pro/Ultra unlimited covers selected image models only |
| Watermark on the file | Free tier: watermark and public by default. Paid: no watermark, private gens | No watermark on kept takes; commercial use included |
| Free-tier grant | 150 tokens/day, no rollover; watermarked, public by default, slower/queued | up to 150 credits once (our algorithm may grant them), no card. Previews free; 5s 720p keep = 105 credits ($1.05) |
Elyum: 1 credit = $0.01, charged only when you keep a take. Leonardo.Ai figures from its own pricing pages and independent reviews, verified 2026-09-07; sources at the end. Full rate card on the Veo 3.1 page.
Where the two differ
Leonardo.Ai uses Fast Tokens on the web app and a separate dollar API. Elyum uses one credit = $0.01. Leonardo.Ai does not publish Veo 3.1 tokens per second. Elyum publishes 20.8 / 28.8 / 60 credits per second at 720p / 1080p / 4K.
Leonardo.Ai deducts tokens on generate. Support says failed jobs should restore tokens, not necessarily instantly; users report emailing [email protected] with generation ID and timestamp. On Elyum, previews are free, kills refund in full, and failed generations are never charged (Free plan: one kill per month).
Elyum Veo 3.1: 4 to 8 second clips at 720p, 1080p, or 4K. Leonardo.Ai offers Veo 3.1 plus Fast/Lite variants; max resolution and clip length for Veo on Leonardo.Ai were not independently verified. Wan 3.0 on Leonardo.Ai is marketed up to 30 seconds (a different model).
Leonardo.Ai free outputs are public and watermarked, with a non-exclusive license and broad Leonardo reuse rights; paid plans get private gens, no watermark, and full IP. Elyum kept takes are unmarked and commercial. Elyum has no team seats; Leonardo.Ai sells Team Starter ($72/mo, 3 seats) and Team Growth.
Leonardo.Ai ships canvas/editing tools, character consistency, and team workspaces. Elyum does not have a timeline editor, stock footage, avatars, a mobile app, team seats, or a community feed. For the fuller desk comparison see the Leonardo.Ai alternative page.
Leonardo.Ai API getting-started docs describe dedicated API credits, a bearer-token key, and dollar billing, distinct from Fast Token web plans. Elyum uses one credit balance in the web studio and from Claude, ChatGPT, or Cursor via MCP and REST.
Pick Leonardo.Ai when
- Veo 3.1 Fast and Lite variants sit next to Lucid Origin, Phoenix, and canvas tools in one subscription.
- Team Starter ($72/mo, three seats) and Team Growth share a token pool; Elyum has no team seats.
- Paid plans include a rollover token bank, so unused monthly Fast Tokens are not fully lost.
- Trustpilot reviewers often praise support speed. Relaxed (token-free) generation covers first-party models on Premium and Ultimate, not Veo.
Pick Elyum when
- Published per-second prices: a kept 5s 720p clip is 105 credits ($1.05). Leonardo.Ai does not publish a Veo 3.1 clip price.
- Previews are free. Credits charge only on KEEP. Killed takes refund in full. Failed jobs are never charged.
- 720p, 1080p, and 4K are listed. Kept downloads have no watermark and include commercial use.
- One credit balance in the web studio, MCP, and REST API. Free plan is up to 150 credits once (our algorithm may grant them), no card.
Veo 3.1 in three styles, rendered on Elyum
Getting the most out of Veo 3.1
- Put spoken lines in quotation marks and name the speaker plus delivery (whisper, close-mic, roomy). Unquoted speech often becomes mutter or burned-in text.
- Direct one camera move per clip (slow dolly-in, locked-off, handheld). Two moves in an 8-second prompt fight each other and smear the last frames.
- For image-to-video, describe only motion, wind, and eye-line from the still. Restyling hair, wardrobe, or lens in the prompt is how faces drift.
- Write sound as a second sentence: room tone, Foley, off-screen traffic. If you skip audio, Veo 3.1 still invents a bed, often the wrong one.
- Keep the story to one beat in 4 to 8 seconds. A prompt that stacks enter, notice the package, then phone a sister is three scenes. The model compresses them into mush.
- State aspect ratio in the prompt (16:9 spot, 9:16 story). Do not rely on negative prompts (no jitter, no morph) as the only control.
Frequently asked questions
How much does a Veo 3.1 video cost on Elyum vs Leonardo.Ai?
Is Veo 3.1 the same model on both?
What happens to a bad Veo 3.1 take?
Do I need a subscription to use Veo 3.1 on Elyum?
Does Ultimate make Veo 3.1 unlimited on Leonardo.Ai?
- Leonardo.Ai Pricing (search-engine summary; direct fetch returned HTTP 403) (accessed 2026-09-07)
- Leonardo.AI pricing 2026: plans, tokens, and the real cost — eesel AI (accessed 2026-09-07)
- AI Video Generation Models: What to Use & When — Leonardo.Ai (accessed 2026-09-07)
- Veo 3 AI Video Generator – Try Now — Leonardo.Ai (accessed 2026-09-07)
- video not generated and credits gone — Adobe Community bug report (accessed 2026-09-07)
- Leonardo.Ai Reviews — Trustpilot (accessed 2026-09-07)
- Leonardo.Ai API getting-started docs (authentication and separate API billing) (accessed 2026-09-07)
- Can You Use AI-Generated Images Commercially? — Leonardo.Ai (accessed 2026-09-07)
Related comparisons
Create everything. Keep the best.
Veo 3.1 is a AI video generator you can preview free on Elyum. Keep the takes you love, kill the rest for a full refund.