OpenAI Launches GPT-6 Astra for Enterprise Automation

2026-09-05 companies

Redmond, Saturday, 5 September 2026.
OpenAI and Microsoft launched GPT-6 Astra, an advanced AI capable of autonomously operating computer interfaces and solving complex scientific tasks, setting a new benchmark for corporate productivity.

Enterprise AI Integration Reaches New Milestone

OpenAI and Microsoft announced the general availability of GPT-6 Astra within Microsoft Foundry on 2026-09-02, marking a significant advancement in enterprise artificial intelligence deployment [1]. The model represents a transition from conversational AI assistance to agentic workflows capable of deliberate planning, multi-step reasoning, and active execution across business applications [1]. GPT-6 Astra features advanced computer-use capabilities that allow it to interact with software interfaces by interpreting on-screen information to perform tasks like updating records, navigating development tools, and testing software [1]. Security for these capabilities is managed via Microsoft Foundry, which enforces scoped credentials, approved resources, mandatory human checkpoints for consequential actions, and activity monitoring [1].

Model Performance and Benchmark Achievements

OpenAI officially released GPT-6 Astra on 2026-09-04, featuring state-of-the-art performance in computer use, browsing, software engineering, cybersecurity, and science [2]. The model achieved a 98% score on FrontierMath Tier 4, a 99.9% score on ARC-AGI-3, and a 100% score on ExploitBench [2]. In safety evaluations, GPT-6 Astra demonstrated a 0% rate of exceeding authorized targets in impossible or difficult task scenarios, compared to a 48% rate for the previous GPT-5.6 Sol model [2]. OSWorld 2.0 latency simulations show GPT-6 Astra achieves a 72.6% performance score in approximately 40 minutes per task, representing a 47% reduction in time compared to GPT-5.6 Sol’s 65.7% score in approximately 75 minutes [2].

Rollout Schedule and Availability

GPT-6 Astra began rolling out to a limited set of organizations on 2026-09-03 and over the coming days will become available to all ChatGPT Plus, Pro, Business, and Enterprise users [3]. The model is also accessible through the OpenAI API, Microsoft Azure, and AWS Bedrock [3]. As of 2026-09-05, the model is now available to all Pro, Enterprise, and Business Premium users in ChatGPT and Codex, though it may take a few days to roll out to Plus and Business users [4]. Microsoft confirmed GPT-6 Astra is rolling out to users in Copilot Cowork and Copilot Studio, with availability varying by region and organization [6].

Pricing Structure and Deployment Options

OpenAI set standard API pricing for GPT-6 Astra at $10 per million input tokens and $50 per million output tokens, with a Fast mode offering 2x processing speed at 2x the standard price [2]. Microsoft Foundry offers GPT-6 Astra in both Global and US Data Zone geographies with Standard pay-as-you-go and Provisioned Throughput deployment options [1]. US Data Zone pricing reflects a 10% premium over Global rates [1]. The model supports a total context window of 1,050,000 tokens with an output limit of 128,000 tokens [7].

Competitive Positioning and Market Analysis

In comparative benchmarks, Claude Fable 5.1 outperforms GPT-6 Astra on Humanity’s Last Exam with tools and the Artificial Analysis Intelligence Index [8]. Artificial Analysis notes that GPT-6 Astra scores equal to GPT-5.6 Sol in the Intelligence Index at 61, which is 5 points lower than Claude Fable 5.1 [5]. However, GPT-6 Astra leads the Coding Agent Index cost efficiency frontier, costing about the same as GPT-5.6 Sol while scoring 2 points higher on the Index [5]. Per task, the model is less than half the cost of Claude Fable 5 for the same score [5].

Enterprise Adoption and Strategic Implications

Albertsons Companies noted that Azure OpenAI on Microsoft Foundry helps create a balance of speed and control so teams can stay focused on delivering meaningful outcomes for customers, associates, and the business [1]. Replit’s CTO Luis Hector Chavez stated that GPT-6 Astra available through Microsoft Foundry unlocks a new level of agentic capability that goes beyond code generation to active software creation [1]. OpenAI reports that GPT-6 Astra is the first model to reach the Critical cybersecurity threshold under its Preparedness Framework, capable of identifying unknown security flaws and developing exploit paths without human intervention [8]. The model demonstrated competitive performance across scientific benchmarks including Terminal-Bench Science, GeneBench Pro, LifeSciBench, MedChemBench, and HealthBench Professional [8].

Sources

Enterprise AI Productivity Tech