clapper / src /services /metrics /constants.ts
jbilcke-hf's picture
jbilcke-hf HF staff
add linter and prettifier
c1f12bf
raw
history blame
4.35 kB
// some providers don't provide APIs to get those statistics
// so we have to resort to using very approximative estimations
import { ComputeProvider } from '@aitube/clapper-services'
import {
ProviderMetricsEstimationType,
ProviderMetricsModelEstimations,
} from './types'
// also note that duration depends on the parameters and image size
// so generally it only make sense to use the averageCostPerSecond
//
// duration is also impact by how many requests we are running at the same time,
// provider load etc.. so sometimes there is a big delta between what
// is announced on the corporate websites and the reality
export const estimatedMetrics: Record<
ComputeProvider,
Record<string, ProviderMetricsModelEstimations>
> = {
[ComputeProvider.NONE]: {
// TODO list the most popular models
},
[ComputeProvider.CUSTOM]: {
// TODO list the most popular models
},
[ComputeProvider.HUGGINGFACE]: {
// TODO list the most popular models
},
[ComputeProvider.COMFY_HUGGINGFACE]: {
// TODO list the most popular models
},
[ComputeProvider.REPLICATE]: {
// TODO list the most popular models
},
[ComputeProvider.COMFY_REPLICATE]: {
// TODO list the most popular models
},
[ComputeProvider.COMFY_COMFYICU]: {
// TODO list the most popular models
},
[ComputeProvider.ELEVENLABS]: {
// TODO list the most popular models
},
[ComputeProvider.OPENAI]: {
// TODO list the most popular models
},
[ComputeProvider.STABILITYAI]: {
// TODO list the most popular models
},
[ComputeProvider.FIREWORKSAI]: {},
[ComputeProvider.GROQ]: {
// TODO list the most popular models
},
[ComputeProvider.KITSAI]: {
// TODO list the most popular models
},
[ComputeProvider.ANTHROPIC]: {
// TODO list the most popular models
},
[ComputeProvider.GOOGLE]: {
// TODO list the most popular models
},
[ComputeProvider.MISTRALAI]: {
// TODO list the most popular models
},
[ComputeProvider.COHERE]: {
// TODO list the most popular models
},
[ComputeProvider.FALAI]: {
'fal-ai/metavoice-v1': {
estimationType: ProviderMetricsEstimationType.MANUAL_MEASUREMENTS,
averageCostPerComputeTimeInSec: 0.0006666666666666666,
averageDurationInSec: 15,
averageCostPerGeneration: 0.01,
},
'fal-ai/fast-sdxl': {
estimationType: ProviderMetricsEstimationType.MANUAL_MEASUREMENTS,
averageCostPerComputeTimeInSec: 0.001105904404873477,
averageDurationInSec: 1.73114,
averageCostPerGeneration: 0.0022138836772983114,
},
'fal-ai/stable-audio': {
estimationType: ProviderMetricsEstimationType.MANUAL_MEASUREMENTS,
averageCostPerComputeTimeInSec: 0.0010714285714285715,
averageDurationInSec: 6.64838,
averageCostPerGeneration: 0.0075,
},
'fal-ai/stable-video': {
estimationType: ProviderMetricsEstimationType.MANUAL_MEASUREMENTS,
averageCostPerComputeTimeInSec: 0.000029239766081871346,
averageDurationInSec: 55.61273333333333,
averageCostPerGeneration: 0.0010416666666666667,
},
'fal-ai/omni-zero': {
estimationType: ProviderMetricsEstimationType.MANUAL_MEASUREMENTS,
averageCostPerComputeTimeInSec: 0.0011475409836065576,
averageDurationInSec: 8.7341,
averageCostPerGeneration: 0.01,
},
'fal-ai/ccsr': {
estimationType: ProviderMetricsEstimationType.MANUAL_MEASUREMENTS,
averageCostPerComputeTimeInSec: 0.0011111111111111111,
averageDurationInSec: 30,
averageCostPerGeneration: 0.015,
},
},
[ComputeProvider.MODELSLAB]: {
// TODO list the most popular models
},
[ComputeProvider.MIDJOURNEY]: {
// TODO list the most popular models
},
[ComputeProvider.SUNO]: {
// TODO list the most popular models
},
[ComputeProvider.UDIO]: {
// TODO list the most popular models
},
[ComputeProvider.LUMALABS]: {
// TODO list the most popular models
},
[ComputeProvider.KUAISHOU]: {
// TODO list the most popular models
},
[ComputeProvider.RUNWAYML]: {
// TODO list the most popular models
},
[ComputeProvider.HEDRA]: {
// TODO list the most popular models
},
[ComputeProvider.LEONARDOAI]: {
// TODO list the most popular models
},
[ComputeProvider.EVERARTAI]: {
// TODO list the most popular models
},
}