Model API usage
Prepaid credits draw down against the applicable API rate. Creative Agent bills the underlying model calls, with no separate agent fee in this offer.
Pricing
API pricing
and usage estimates
AGIC model APIs use consumption-based billing. Explore public rate references, estimate a workload and choose the right deployment.
Open rate references
Public provider reference rates
Public provider rates
These are published provider prices for planning, not a Mlytics quotation. The reference models below do not imply Mlytics availability or a provider partnership. Mlytics-supported routes and commercial rates are confirmed with API access.
| Capability / model | Reference provider | Public rate | Configuration | Source |
|---|---|---|---|---|
| Text & reasoningQwen3.5-9B | Together AIProvider-hosted reference |
$0.17Input / 1M tokens
$0.25Output / 1M tokens
|
Serverless FP8 route. Standard input / output; no caching or batch discount assumed. | Source: Together public pricing for Qwen3.5-9B |
| Video generationWan2.2-T2V-A14B | DeepInfraProvider-hosted reference |
$0.033Output / generated second
|
720p · 16 fps · fixed 5-second clips. $0.165 per clip. Landscape or portrait. | Source: DeepInfra model pricing for Wan2.2-T2V-A14B |
| Image generationFLUX.1 [schnell] | DeepInfraProvider-hosted reference |
$0.002Configured output / image
|
1024 × 1024 · 4 inference steps. Calculated from the provider formula: $0.0005 × (width/1024) × (height/1024) × steps. | Source: DeepInfra pricing formula for FLUX.1 [schnell] |
| Text to speechKokoro-82M | Together AIProvider-hosted reference |
$4Input text / 1M characters
|
Character-based speech synthesis. Language and voice selection follow the provider model. | Source: Together public pricing for Kokoro-82M |
| Speech to textWhisper large-v3 | Together AIProvider-hosted reference |
$0.0015Input audio / audio minute
|
Non-streaming transcription. Input audio duration is the usage meter. | Source: Together transcription pricing for Whisper large-v3 |
Showing all 5 reference models.
List prices can change. No promotional, caching, batch or volume discounts are assumed. Delivery, storage, taxes, support, integration work and AI Mode media are excluded.
Cost explorer
Estimate your workload
Adjust the quantities to see the model-usage subtotal at the reference rates above. The calculator stays in your browser.
How the commercial layers fit
Generation, delivery and media
Content distribution
Multi-CDN delivery is itemised separately. Rates depend on the traffic profile, regions and service configuration. Request a delivery scope.
AI Mode campaigns
Media and brand partnerships have their own scope, inventory and commercial agreement. API credits do not include campaign placement.
For steady volume or specific requirements
Explore Enterprise
Need reserved capacity?
Explore dedicated compute, contracted usage and a deployment scoped to your business.
Questions
Pricing questions
Are these Mlytics selling prices?
The reference rate card uses the named providers’ public prices as transparent references. It is not an approved Mlytics quote, a promise of model availability, or a statement of partnership. Your Mlytics catalogue, rates and service scope are confirmed before production.
Why do voice services use different units?
The underlying services meter different work. This reference card bills Kokoro by input characters and Whisper by input audio minutes. There is no assumed conversion between characters, tokens and audio duration.
How is video usage calculated?
This reference uses DeepInfra’s Wan2.2 route at $0.033 per generated second. It generates fixed 5-second, 720p clips, so one clip is $0.165 and three clips are $0.495 before any additional services.
How is the image reference calculated?
DeepInfra publishes $0.0005 × (width/1024) × (height/1024) × inference steps. At 1024 × 1024 and four steps, that is $0.002 per image. Other settings change the cost.
Does the estimate include a complete campaign?
It includes only the model quantities you enter at the reference rates. Delivery, storage, media, taxes, support, integration work and any additional model calls are outside the subtotal.
Can we agree enterprise terms?
Yes. Reserved capacity and eligible API usage can be scoped under a monthly commitment, with contract rates, usage drawdown and agreed overages. See the Enterprise page or contact Mlytics.
Confirm your AGIC rates and model access
Bring your model requirements and expected usage. We’ll confirm the available routes and commercial terms.