specification: API Commons Plans
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/Plans
provider: Hyperbolic
providerId: hyperbolic-ai
created: '2026-05-25'
modified: '2026-05-25'
reconciled: true
tags:
  - AI
  - GPU
  - Inference
  - Marketplace
description: 'Reconciled plans and per-token / per-image / per-GPU-hour pricing for the Hyperbolic Serverless
  Inference and GPU Marketplace services. Source: hyperbolic.ai, docs.hyperbolic.ai, and third-party
  pricing trackers as of May 2026.'
sources:
  - https://www.hyperbolic.ai/
  - https://docs.hyperbolic.ai/inference/overview
  - https://docs.hyperbolic.ai/on-demand/overview
  - https://huggingface.co/docs/inference-providers/en/providers/hyperbolic
plans:
  - id: hyperbolic-basic
    name: Basic
    type: tier
    description: Free tier — 60 requests per minute, community support, no minimum deposit.
    entries:
      - label: Request rate
        name: rpm
        type: metered
        metric: requests
        limit: 60
        timeFrame: minute
        geo: global
        unit: 1
        price: '0.00'
        userMultiplied: false
    elements:
      - name: 60 requests/minute
      - name: Community support
      - name: No minimum deposit
  - id: hyperbolic-pro
    name: Pro
    type: tier
    description: Pro tier — 600 requests per minute, priority and email support, requires at least $5 deposit.
    entries:
      - label: Request rate
        name: rpm
        type: metered
        metric: requests
        limit: 600
        timeFrame: minute
        geo: global
        unit: 1
        price: '0.00'
        userMultiplied: false
      - label: Minimum deposit
        name: deposit_minimum
        type: prepaid
        metric: usd
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1
        price: '5.00'
        userMultiplied: false
    elements:
      - name: 600 requests/minute
      - name: Priority + email support
      - name: $5 minimum deposit
  - id: hyperbolic-enterprise
    name: Enterprise
    type: tier
    description: Enterprise tier — unlimited rate limits, 24/7 dedicated support, custom contracts.
    entries:
      - label: Request rate
        name: rpm
        type: metered
        metric: requests
        limit: -1
        timeFrame: minute
        geo: global
        unit: 1
        price: 'Call'
        userMultiplied: false
    elements:
      - name: Unlimited rate limits
      - name: 24/7 dedicated support
      - name: Custom contracts and SLA
  - id: hyperbolic-inference-llama-405b
    name: Llama 3.1 405B Inference
    type: usage-based
    description: Per-million-token pricing for Meta-Llama-3.1-405B-Instruct.
    entries:
      - label: Tokens (input + output)
        name: tokens
        type: metered
        metric: tokens
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1000000
        price: '4.00'
        userMultiplied: false
  - id: hyperbolic-inference-deepseek-r1
    name: DeepSeek R1 Inference
    type: usage-based
    description: Per-million-token pricing for DeepSeek-R1.
    entries:
      - label: Tokens (input + output)
        name: tokens
        type: metered
        metric: tokens
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1000000
        price: '3.00'
        userMultiplied: false
  - id: hyperbolic-inference-llama-70b
    name: Llama 3.3 70B Inference
    type: usage-based
    description: Per-million-token pricing for Llama-3.3-70B and similar 70B-class models.
    entries:
      - label: Tokens (input + output)
        name: tokens
        type: metered
        metric: tokens
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1000000
        price: '0.40'
        userMultiplied: false
  - id: hyperbolic-inference-small-llms
    name: Small LLMs Inference
    type: usage-based
    description: Per-million-token pricing floor for 7B–8B class models.
    entries:
      - label: Tokens (input + output)
        name: tokens
        type: metered
        metric: tokens
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1000000
        price: '0.10'
        userMultiplied: false
  - id: hyperbolic-inference-vision
    name: Vision Models Inference
    type: usage-based
    description: Per-million-token pricing floor for multimodal vision models (Llama 3.2 Vision, Qwen2-VL).
    entries:
      - label: Tokens (input + output)
        name: tokens
        type: metered
        metric: tokens
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1000000
        price: '0.15'
        userMultiplied: false
  - id: hyperbolic-image-generation
    name: Image Generation
    type: usage-based
    description: Per-image pricing floor for SDXL / SD3.5 / FLUX diffusion models.
    entries:
      - label: Images generated
        name: images
        type: metered
        metric: image
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1
        price: '0.0025'
        userMultiplied: false
  - id: hyperbolic-audio-tts
    name: Audio TTS
    type: usage-based
    description: Per-1000-character pricing floor for text-to-speech.
    entries:
      - label: Characters synthesized
        name: characters
        type: metered
        metric: character
        limit: -1
        timeFrame: usage
        geo: global
        unit: 1000
        price: '0.001'
        userMultiplied: false
  - id: hyperbolic-gpu-rtx4090
    name: GPU On-Demand — RTX 4090
    type: usage-based
    description: Per-GPU-hour pricing for RTX 4090 on-demand rental.
    entries:
      - label: GPU hours
        name: gpu_hours
        type: metered
        metric: gpu_hour
        limit: -1
        timeFrame: hour
        geo: global
        unit: 1
        price: '0.50'
        userMultiplied: false
  - id: hyperbolic-gpu-a100
    name: GPU On-Demand — A100
    type: usage-based
    description: Per-GPU-hour pricing for NVIDIA A100.
    entries:
      - label: GPU hours
        name: gpu_hours
        type: metered
        metric: gpu_hour
        limit: -1
        timeFrame: hour
        geo: global
        unit: 1
        price: '1.80'
        userMultiplied: false
  - id: hyperbolic-gpu-h100
    name: GPU On-Demand — H100
    type: usage-based
    description: Per-GPU-hour pricing for NVIDIA H100. Marketplace pricing ranges from $1.39 to $3.20 per
      hour depending on supplier, configuration, and dedicated vs shared.
    entries:
      - label: GPU hours (entry marketplace)
        name: gpu_hours_marketplace
        type: metered
        metric: gpu_hour
        limit: -1
        timeFrame: hour
        geo: global
        unit: 1
        price: '1.39'
        userMultiplied: false
      - label: GPU hours (dedicated)
        name: gpu_hours_dedicated
        type: metered
        metric: gpu_hour
        limit: -1
        timeFrame: hour
        geo: global
        unit: 1
        price: '3.20'
        userMultiplied: false
  - id: hyperbolic-gpu-h200
    name: GPU On-Demand — H200
    type: usage-based
    description: Per-GPU-hour pricing for NVIDIA H200 on-demand rental.
    entries:
      - label: GPU hours
        name: gpu_hours
        type: metered
        metric: gpu_hour
        limit: -1
        timeFrame: hour
        geo: global
        unit: 1
        price: 'Marketplace'
        userMultiplied: false
  - id: hyperbolic-reserved-clusters
    name: Reserved Clusters
    type: commitment
    description: 3-12 month prepaid GPU cluster commitments with volume discounts up to 40% off marketplace
      on-demand pricing.
    entries:
      - label: Volume discount
        name: discount
        type: discount
        metric: percent
        limit: 40
        timeFrame: usage
        geo: global
        unit: 1
        price: '-40'
        userMultiplied: false
  - id: hyperbolic-marketplace-fee
    name: Marketplace Platform Fee
    type: commission
    description: Hyperbolic charges a 10% platform fee on rental income from GPU suppliers.
    entries:
      - label: Platform fee
        name: platform_fee
        type: commission
        metric: percent
        limit: 10
        timeFrame: usage
        geo: global
        unit: 1
        price: '10'
        userMultiplied: false