Agent S
https://arxiv.org/abs/2410.08164
Open agentic framework that uses computers like a human.
Active · Verified 2026-10-28 · fresh as of 2026-10-28 · 24 cited facts, 0 with a third-party source · Suggest a correction
At a glance
- Type
- Open agentic computer-use framework vendor fix
- Tier
- T4
- Latest release
- v0.3.2 2025-12-16 vendor fix
- License
- Apache-2.0 repo fix
- GitHub
- 11.1k★ +545/mo Python repo fix
- Pricing
- not applicable — research paper fix
- Funding
- not applicable — not commercial fix
- Freshness
- 2026-10-28 fix
- Layer
- agent-infra repo fix
- Licensing
- open-source repo fix
Memory taxonomy
- storage
- hybrid
- retrieval
- similarity
- persistence
- long-term
- update
- extraction
- unit
- episode
- governance
- inspectable
- conflict
- llm-arbitrate
When to use
Optimised for: not applicable - research paper
Anti-fit: not applicable - research paper
Pros & cons
Pros
Computer-use agent with explicit screen-history memory.
Cons
Preprint; screen-history scales aggressively.
Claims & capabilities
Open agentic framework using computers like a human via experience-augmented hierarchical planning + Agent-Computer Interface (ACI). Headline: 20.58% success rate vs. 11.21% baseline on OSWorld (83.6% relative improvement, +9.37pp absolute); baseline: GPT-4o vanilla OSWorld agent; primary datasets: OSWorld (369 Ubuntu tasks) and WindowsAgentArena.
Technical surface
- API surface
- not applicable — research paper
- Backend storage
- not applicable — research paper
- Deployment
- not applicable — research paper
- Embedding model
- not applicable — research paper
- Multi-tenancy
- not applicable — research paper
- MCP
- not applicable — research paper, no deployed product
- A2A
- not applicable — research paper, no deployed product
- OpenTelemetry
- not applicable — research paper, no deployed product
Similar systems
Other recent method papers — theorized, no distinct product in the catalog, ranked by inbound references.
- Compressive Transformer T3
Maintains recent states in full resolution while compressing older memories with learned compression functions. DeepMind.
- MemGPT v2 / agent-tools T3
Already in catalog as the foundational MemGPT paper. Note: Letta is the productionised successor (cross-listed).
- Transformer-XL T3
Extends context through segment-level recurrence + caching of hidden states from prior segments. Foundational long-context architecture.
- Generative Agents T3
Park et al. — landmark agent-simulation paper. Reflection + memory stream + retrieval enable believable agent behavior.
- MemoryBank T3
Enhances LLMs with long-term memory. Early influential paper.
- Reflexion T3
Language agents with verbal reinforcement learning.
Sources (23 cited facts, 0 third-party)
Each fact links to the page it came from; the badge says who published it (how to read sources).
- What it is Open agentic computer-use framework vendor
- What it does & what's distinct Open agentic framework that uses computers like a human. vendor
- Claims / benefits Open agentic framework using computers like a human via experience-augmented hierarchica… vendor
- Modalities text + image (GUI — multimodal desktop/web via screenshots) vendor
- Created 2024-10 vendor
- Latest release v0.3.2 2025-12-16 vendor
- License Apache-2.0 repo
- GitHub 11.1k★ +545/mo Python repo
- Mindshare 147 cites · 7.8/mo · 11142* / 1301 forks · 10 influential SS cites paper
- Citations 146 cites 73/yr paper
- Founders / pedigree Simular Research; no institutional affiliation confirmed vendor
- Memory volume / scale no scale claim — research (OSWorld benchmark) vendor
- Performance 20.58% OSWorld success +9.37pp OSWorld vs GPT-4o vendor
- Reproducibility Original team only — 1301 forks of https://github.com/simular-ai/Agent-S (mostly fork-on… repo
- Code/weights release Code public + weights closed — https://github.com/simular-ai/Agent-S · 11142* / 1301 for… repo
- Contradiction handling LLM resolves and overwrites vendor
- Pros Computer-use agent with explicit screen-history memory. vendor
- Cons Preprint; screen-history scales aggressively. vendor
- Project & sources arxiv 2410.08164 vendor
- commit-trajectory [{"ym":"2024-10","cum":36},{"ym":"2024-11","cum":85},{"ym":"2024-12","cum":88},{"ym":"20… repo
- citation-trajectory [{"year":2025,"cum":2,"infl":0}] paper
- licensing-bucket open-source repo
- taxonomy-three-layer-rollup agent-infra repo
Related systems
References (1)
- Voyager cites — S2 isInfluential citation