Replicate Pricing Calculator
Updated May 2026Pure-usage per-second public-model + per-output image + per-token LLM + per-second dedicated GPU + sales-led enterprise
Replicate pricing: Replicate pricing 2026: time-billed public models, FLUX 1.1 Pro $0.04/img, DeepSeek R1 $3.75/1M, dedicated H100 $0.001525/sec, A100 $0.0014/sec, idle-time-free. Use the free calculator below to enter your usage and get an instant all-in monthly estimate — including overages — and see which plan is cheapest for your needs.
How Replicate prices: Pure-usage per-second public-model + per-output image + per-token LLM + per-second dedicated GPU + sales-led enterprise.
How Replicate prices
Pure-usage per-second public-model + per-output image + per-token LLM + per-second dedicated GPU + sales-led enterprise
Your usage
💡 A 3-sec FLUX Dev run = 3 GPU-sec; 10k runs/day ≈ 900k GPU-sec/mo
Public models bill at the per-second rate of the chosen hardware; dedicated deployments use the same rates. Idle time is free on public models and fast-booting fine-tunes.
💡 Premium image models bill per output image instead of per second
Select per-output image models price each generated image; per-second time billing is often cheaper if you can predict inference latency.
💡 1M input tokens ≈ 750k words of prompts/context
Per-token LLMs are billed per million input + output tokens.
💡 Output tokens are the model's generated response
Output tokens are priced higher than input; match this to the same model you picked for the input rate.
Monthly Estimate
Pay-as-you-go
$4,575.00
per month
Cost Breakdown
Annual
$54,900.00
Why Pay-as-you-go — $4,575.00/mo
- Pay-as-you-go at $4,575.00/mo.
- How we got there: 2M GPU-seconds (Hardware compute time) × 0 $/sec (Hardware tier (per-second rate)) = 2.8K $.
- How we got there: 50K images (Per-output images) × 0 $/image (Image model (per-output rate)) = 1.3K $.
- How we got there: 50 M tokens (LLM input tokens) × 3 $/1M in (LLM model (input rate)) = 150 $.
- How we got there: 25 M tokens (LLM output tokens) × 15 $/1M out (LLM model (output rate)) = 375 $.
- Hardware compute time (2.8K $ × $1) is 61% of your bill — the lever that matters most.
- Need less? Hardware tier (per-second rate): CPU Small — $0.000025/sec drops you to Pay-as-you-go at $1,825.00/mo.
Budget range
Plan for usage swings, not just today's estimate.
Conservative
$3,175.00
1M GPU-seconds/mo
Expected
$4,575.00
2M GPU-seconds/mo
Aggressive
$7,375.00
4M GPU-seconds/mo
Need a calculator like this on your pricing page?
Embed interactive pricing calculators on your website to help customers understand costs and boost conversions.
Get StartedAbout this Replicate calculator
This calculator estimates your Replicate cost from publicly available pricing. Actual costs may vary with your specific agreement, volume discounts, and usage patterns — always verify on the provider's official pricing page for the most current rates.
Replicate pricing — frequently asked questions
How much does Replicate cost per month? ▼
Replicate has no monthly subscription fee — you pay only for the per-second execution time of public and dedicated models, plus any per-output or per-token pricing for premium models. A typical FLUX Dev workload generating 1,000 images/month would cost $25 (at $0.025/image) or roughly $5–$8 per-second time-billed depending on inference latency.
What is the difference between Replicate public models and dedicated deployments? ▼
Public models are multi-tenant — they share GPU capacity with other customers and billing is per-second of execution time on the underlying hardware. Dedicated deployments are single-tenant: your own GPU instance with guaranteed throughput, billed per-second only when active (no idle charges). Public models are cheaper for sporadic workloads; dedicated wins at sustained high QPS.
What are Replicate's per-second GPU rates? ▼
Dedicated deployment per-second rates: CPU Small $0.000025, CPU $0.0001, T4 $0.000225, L40S $0.000975, A100 80GB $0.0014, H100 $0.001525. Translated to per-hour: A100 ~$5.04/hr, H100 ~$5.49/hr. Multi-GPU deployments available with committed spend contracts.
Does Replicate have per-output or per-token pricing? ▼
Yes for select premium models. Per-image: FLUX 1.1 Pro $0.04, FLUX Dev $0.025, Ideogram v3 $0.09. Per-token: Claude 3.7 Sonnet $3.00 input / $15 output per 1M, DeepSeek R1 $3.75 input / $10 output per 1M. Most public models default to per-second time billing.
Is this Replicate pricing calculator free and accurate? ▼
Yes — it's completely free with no signup. It uses the latest publicly available Replicate pricing (verified May 2026). Actual costs may vary with volume discounts or enterprise terms — always confirm on Replicate's official pricing page.