OpenAI: GPT-6 Astra在线

GPT-6 Astra is OpenAI’s flagship AI model for complex reasoning, coding, computer use, research, science, and professional knowledge work. It is designed for long-running agentic workflows that require tool use, multistep planning, iteration, and adaptation to changing instructions. On 开桥, developers can access the model using the model ID openai/gpt-6-astra.

OpenAI released GPT-6 Astra on September 3, 2026. The GPT-6 Astra model is now available on 开桥 through the unified API using the model ID openai/gpt-6-astra.

GPT-6 Astra is best suited to demanding AI and agentic workflows, including:

• Software engineering, debugging, refactoring, and code review
• Browser and computer-use automation
• Data analysis, spreadsheets, and business intelligence
• Scientific and technical research
• Document creation and professional knowledge work
• Secure code review and vulnerability patching

It is particularly useful when an application must work across code, browsers, documents, data, and other tools rather than produce only a single text response.

开桥 currently lists GPT-6 Astra input pricing at $10–$20 per 1M tokens and output pricing at $50–$75 per 1M tokens, based on the pricing conditions shown for the available routes. Cache reads are listed at $1–$2 per 1M tokens, while cache writes are listed at $12.50–$25 per 1M tokens.

Developers can view current pricing, latency, throughput, cache performance, and availability directly on this page. Because these figures may change, use the live route data when estimating production costs.

GPT-6 Astra supports a 1.05M-token context window on the routes currently listed by 开桥. This large context capacity makes the model suitable for analyzing extensive documents, reviewing large codebases, maintaining long agent histories, and processing complex research materials.

Create a 开桥 API key, use the base URL https://openqiao.com/api/v1, and specify openai/gpt-6-astra in the request’s model field. 开桥 supports both Chat Completions and Responses API formats, making it easier to integrate GPT-6 Astra into applications that already use compatible SDKs. See the 开桥 API Quickstart for request examples and setup instructions.

Yes. GPT-6 Astra is designed for agentic workflows involving coding environments, browsers, professional software, data, documents, and other tools. It can plan a task, use available tools, inspect the results, troubleshoot failures, and continue working toward a defined goal. The model is also designed to incorporate new requirements while retaining earlier instructions, which is useful for long-running and changing workflows.

OpenAI reports that GPT-6 Astra achieved 98% on FrontierMath Tier 4, 99.9% on ARC-AGI-3, 57.9% on Terminal-Bench 4.0, 59.3% on Agents’ Last Exam, 72.6% on OSWorld 2.0, and 41.4% on AutomationBench. These are vendor-reported benchmark results obtained under specific evaluation settings. Developers should also test the model with their own prompts, tools, datasets, and production requirements.

开桥 provides access to GPT-6 Astra through a unified API while displaying route-level pricing, latency, throughput, cache performance, uptime, and availability. This visibility helps developers evaluate routes and control costs without maintaining separate API integrations. A single 开桥 API key can also provide access to other available AI models, making it easier to test and manage different workloads through one platform.