generated: '2026-07-21'
method: searched
source: https://thetokencompany.com/blog ; https://thetokencompany.com/llms.txt
scheme: model-generation
current_version: bear-2
note: >-
  The Token Company does not publish a dated API changelog; product evolution is
  announced as model releases on the blog. Entries below are the model-release
  milestones surfaced on the blog / llms.txt.
entries:
  - version: bear-2
    type: model-release
    breaking: false
    highlights: Latest, recommended compression model; default across SDKs.
  - version: bear-1.2
    type: model-release
    breaking: false
    highlights: Previous-generation compression model.
  - version: bear-1.1
    type: model-release
    breaking: false
    url: https://thetokencompany.com/blog/bear-1-1
    highlights: Improved accuracy preservation and faster compression.
  - version: bear-1
    type: model-release
    breaking: false
    url: https://thetokencompany.com/blog/bear-1
    highlights: First LLM input compression model.