specification: FinOps Framework
specificationVersion: '1.0'
schema: https://www.finops.org/framework/
provider: Tabby
providerId: tabby-ml
created: '2026-07-11'
modified: '2026-07-11'
reconciled: false
tags:
- AI Coding Assistant
- Code Completion
- Open Source
- FinOps
- Cost Management
- FOCUS
description: >-
  FinOps view of Tabby spend. Tabby's server is open source (Apache-2.0) and free
  to self-host - in that model there is no vendor invoice at all; the cost is your
  own compute for running the Tabby server, dominated by the GPU (or CPU) that
  serves the completion and chat models, plus storage for the indexed knowledge
  base. Because Tabby makes no external calls to third parties, there are no
  per-token inference charges to a model vendor. TabbyML's hosted plans convert
  that owned-infrastructure cost into a predictable per-seat subscription: free
  Community (up to 5 users), Team at 19 USD per seat per month (up to 50 users),
  and custom Enterprise.
notes: >-
  Self-hosted cost is essentially GPU/CPU time and storage, not a Tabby license
  fee. Hosted per-seat rates were captured on 2026-07-11; verify on the pricing
  page. The main optimization lever self-hosted is right-sizing the model and
  device to your team's latency needs.
sources:
- https://www.tabbyml.com/
- https://www.tabbyml.com/pricing
- https://github.com/TabbyML/tabby
- https://focus.finops.org/focus-specification/v1-3/
alignedWith:
  framework: FinOps Foundation Framework
  frameworkUrl: https://www.finops.org/framework/
  dataSpec: FOCUS
  dataSpecVersion: '1.3'
  dataSpecUrl: https://focus.finops.org/focus-specification/v1-3/
publisherName: TabbyML
serviceCategory: AI and Machine Learning
billingModel:
  pricingCategory: Subscription-Based
  billingFrequency: Monthly
  billingCurrency: USD
  chargeCategories:
  - Purchase
  - Usage
focusColumns:
  ServiceName: Tabby
  ServiceCategory: AI and Machine Learning
  ProviderName: TabbyML
  PublisherName: TabbyML
  InvoiceIssuerName: TabbyML
  BillingCurrency: USD
  ChargeCategory: Purchase
  PricingCategory: Subscription-Based
meters:
- name: hosted_seats
  description: Per-seat subscription on hosted Team plans, billed monthly.
  unit: seats
  aggregation: sum
  dimensions:
  - account
- name: self_hosted_compute
  description: GPU/CPU time to run the self-hosted Tabby server (no TabbyML invoice).
  unit: hours
  aggregation: sum
  dimensions:
  - deployment
  - device
- name: index_storage
  description: Storage for the ingested knowledge base and repository index.
  unit: bytes
  aggregation: sum
  dimensions:
  - deployment
principles:
- name: Visibility
  description: On self-hosted, track GPU/CPU hours and storage for the Tabby server; on hosted, track active seats against the plan cap.
- name: Allocation
  description: Map self-hosted Tabby infrastructure to the team(s) it serves, or allocate hosted seats per developer/team.
- name: Optimization
  description: Right-size the model and device to your latency needs; self-host to avoid per-seat fees when you have spare GPU capacity, or use hosted plans to avoid running infrastructure.
- name: Accountability
  description: Assign an owner for the Tabby deployment and review monthly compute/storage cost or seat count against team growth.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com