Agents
7 AI agents & agent frameworks
Agents are classified separately from chatbots and models: systems that plan and execute multi-step tasks with tools. Capabilities are editorial classifications from official documentation; autonomy is described, not scored.
Agent capability matrix
| # | Agent | Type | Autonomy | Browser | Computer | Code exec | Memory | Approval | Surfaces | From |
|---|---|---|---|---|---|---|---|---|---|---|
| 72 | Browser UseBrowser Useunchanged | Browser & computer-use agents | semi-autonomous | ✓ | – | – | – | ✓ | SDK, cloud | OSS |
| 78 | ChatGPT AtlasOpenAI2places up | Browser & computer-use agents | semi-autonomous | ✓ | – | – | ✓ | ✓ | browser | — |
| 92 | CometPerplexity2places up | Browser & computer-use agents | semi-autonomous | ✓ | – | – | – | ✓ | browser | Free |
| 121 | ChatGPT agentOpenAI1places down | General-purpose agents | autonomous | ✓ | ✓ | ✓ | – | ✓ | web | $20/mo |
| 122 | Claude in ChromeAnthropic1places down | Browser & computer-use agents | semi-autonomous | ✓ | – | – | – | ✓ | browser | — |
| 129 | DiaThe Browser Company2places down | Browser & computer-use agents | assistive | ✓ | – | – | – | ✓ | browser | — |
| – | OperatorOpenAIDeprecated | Browser & computer-use agents | semi-autonomous | ✓ | – | – | – | ✓ | web | — |
“–” means the capability is not listed in official documentation, not that it is impossible. Security models and task limits: see each agent's official docs (linked on its profile).