# microsoft/fara

Fara1.5 – A family of frontier computer use agent models

Repository: https://github.com/microsoft/fara
Canonical: https://ross.abutalabs.com/products/fara
Homepage: https://www.microsoft.com/en-us/research/articles/fara1-5-computer-use-agent/
Language: Python
License: MIT
License Family: permissive
Topics: agent, browser-use, computer-use, computer-use-agent, cua
Last push: 2026-07-22T17:03:30+00:00

## Health v2 (maintenance only)
Score: 58/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 93, release rhythm 35, longevity 22
- inputs: {"age_days": 308, "days_push": 42, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 6153, forks 601 (observed 2026-08-28T04:09:36.657387+00:00)

## What it is
Fara1.5 is a family of open-weight computer use agent (CUA) models from Microsoft Research, released at 4B, 9B, and 27B scales and built on Qwen3.5. The repository includes model weights, an evaluation harness with benchmarks like WebTailBench and CUAVerifierBench, and a Universal Verifier for judging agent trajectories.

## Use cases
- automate browser tasks with an AI agent
- run a computer use agent model locally
- evaluate web agent performance on benchmarks
- score agent trajectories with a verifier judge
- download open-weight CUA models from Hugging Face
- benchmark web automation agents on Online-Mind2Web

## When to choose
- you need an open-weight model that can control a browser or computer to complete tasks
- you want state-of-the-art small-model web agent performance with reproducible evaluation tooling
- you are researching computer use agents and need benchmark datasets and verifiers

## When to avoid
- you need a simple RPA tool without GPU hardware or ML infrastructure
- you want a hosted no-code automation product rather than models and evaluation code
- your use case is desktop-only GUI automation outside web browsing

## Facets
- artifact type: library
- maturity: active
- function: agent-framework, machine-learning, llm-inference, benchmarking, web-scraping
- domain: artificial-intelligence, large-language-models, developer-tools
- platform: python, cross-platform
- tags: computer-use-agent, browser-automation, cua, model-weights, webtailbench, qwen, huggingface, ai-agents, automation, gpu

## Member repositories
- microsoft/fara (main) score 58

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:36.657387+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:47:31.357816+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/fara (fetched 2026-08-28T04:09:36.657387+00:00, sha b115bdfb64f4)
  - homepage: https://www.microsoft.com/en-us/research/articles/fara1-5-computer-use-agent/ (fetched 2026-08-29T08:44:31.373141+00:00, sha f80108e5aeee)
  - site_page: https://www.microsoft.com/en-us/research/about-microsoft-research (fetched 2026-08-29T08:44:31.383300+00:00, sha 662089aedfa3)
  - site_page: https://learn.microsoft.com/docs (fetched 2026-08-29T08:44:31.386563+00:00, sha f29800be2b9a)
  - site_page: https://www.microsoft.com/about (fetched 2026-08-29T08:44:31.389068+00:00, sha e30c635c0605)
- Data as of 2026-08-30T08:39:29.467469+00:00.