We may earn a commission if you buy through links on this page. It never changes a score or a ranking. How we make money
Pricing is quote-based.
Replicate has no published monthly or annual plan price: the pricing page lists usage-based billing. Public models charge by runtime or by input/output, while most private models are billed for all time online. Listed examples range from $3 per 1,000 FLUX Schnell images to $0.25 per second of 720p video; private hardware is separately priced by the second or hour. There is no free plan listed, and a free-trial offer is not published. Your bill depends on the model and deployment.
01Every plan, line by line
| Plan | Billed annually | Billed monthly | Built for |
|---|---|---|---|
| Public models — FLUX 1.1 Pro | — | Quote | Text-to-image model users |
| Public models — FLUX Dev | — | Quote | Text-to-image model users |
| Public models — FLUX Schnell | — | Quote | Fast image generation users |
| Public models — DeepSeek R1 | — | Quote | Reasoning model users |
| Public models — Ideogram v3 Quality | — | Quote | High-quality image generation users |
| Public models — Recraft v3 | — | Quote | Text-to-image model users |
| Public models — Wan 2.1 I2V 480p | — | Quote | Image-to-video model users |
| Public models — Wan 2.1 I2V 720p | — | Quote | High-resolution video generation users |
| Private hardware — CPU | — | Quote | Private model deployments |
| Private hardware — base GPUs | — | Quote | Private model deployments |
| Private hardware — 2x GPUs | — | Quote | Higher-capacity private deployments |
| Private hardware — additional A100 GPUs | — | Quote | Multi-GPU private deployments |
| Private hardware — additional H100 GPUs | — | Quote | Multi-GPU private deployments |
| Private hardware — H200 GPUs | — | Quote | H200 private deployments |
| Private hardware — additional L40S GPUs | — | Quote | Multi-GPU private deployments |
| Enterprise / volume discounts | — | Quote | High-volume or complex requirements |
Add-ons and overages
- Additional multi-GPU capacity may require committed-spend contracts; rates are listed with the hardware options.
- Volume discounts are available for large amounts of spend; discount amounts are not published.
02What the plan grid doesn't tell you
Private models bill while online
The pricing page says most private models are billed for all time online, so idle deployment time can count.
GPU rates accrue by time
Private GPU rates are listed per second and per hour; for example, an H100 is $5.49 per hour.
Multi-GPU may require commitment
Replicate says additional multi-GPU capacity may require committed-spend contracts; commitment terms are not listed.
Discounts aren't quantified
Volume discounts are available, but discount amounts and enterprise pricing are not published.
03Which plan should you buy?
| If you are… | Get | Why |
|---|---|---|
| Lower-cost listed image generation | Public models — FLUX Schnell | Costs $3 per 1,000 output images, the lowest listed public-image rate among these examples. |
| Reasoning model API users | Public models — DeepSeek R1 | Pricing is per token: $3.75 per million input tokens and $0.01 per thousand output tokens. |
| Private GPU deployments | Private hardware — base GPUs | Listed single-GPU options include A100, H100, L40S and one thread, priced from $0.81 to $5.49 per hour. |
04Is Replicate worth the money?
Replicate’s appeal is access to many models through one API, with usage-based rates instead of a fixed subscription. That can suit teams that want to choose models and pay for their use, but it makes spend harder to forecast across different billing units. Private deployments add time-based hardware charges, and multi-GPU commitments or enterprise discounts are not fully priced publicly. It may be worth considering for model flexibility; compare expected usage and support requirements before committing.
Replicate pricing FAQ
How much does Replicate cost?
There is no monthly or annual plan price listed. The pricing page shows usage-based rates: examples include FLUX Schnell at $3 per 1,000 output images, FLUX Dev at $0.025 per image, and Wan 2.1 I2V 720p at $0.25 per second of output video.
Does Replicate have a free plan or free trial?
The pricing information indicates that Replicate does not have a free plan. A free trial is not published on the vendor's site, so we would not assume trial credits or a no-cost usage allowance without confirmation from Replicate.
How are private models billed?
The pricing page says most private models are billed for all time online. Private hardware rates are shown per second and per hour; for example, the listed H100 rate is $0.001525 per second or $5.49 per hour.
Are Replicate's volume discounts and enterprise prices public?
No. The pricing information says volume discounts are available, but discount amounts and enterprise pricing are not published. It also lists higher GPU limits and priority support for enterprise or volume customers, without publishing their prices.