Expired50% confidenceBenchmarkExact time
GPT-6 Astra ranks first on agent and computer-operation leaderboards including Terminal Bench 4, OS World, Agents Last Exam, and Automation Bench.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
9/26/2026
First Seen
Valid until: 10/10/2026(expired)
Sources
Related Entities
Related Claims
VerifiedGPT-6 Astra is capable of operating modeling software and game engines, possessing agentic capabilities to view the screen, click on the interface, inspect work, and continue making modifications.74% similarUnverifiedGPT-6 Astra's Computer Use capability is a complete closed loop: the model reads the task and screen state to propose the next action, the runner executes the action in the browser or desktop, and the resulting screenshot, page text, and errors are fed back to the model for decision-making.74% similarUnverifiedAccording to Databricks' internal evaluation, GPT-6 Astra has reached industry-leading levels on enterprise-grade benchmarks including OfficeQA Pro, OfficeQA Pro V2, and Document Processing.72% similarUnverifiedGPT-6 Astra has set a new industry record in computer use capability, achieving SOTA performance.72% similarPartially VerifiedGPT-6 Astra's capability improvements are focused on five key areas: cybersecurity, professional work, software engineering, scientific research, and computer operation.71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/951502API
curl https://kongchang.com/api/v1/knowledge/claims/951502MCP
get_claim(id=951502)