Wednesday, 18 November 2026

OpenAI Launches GPT-6.2 Astra With Stronger Coding, Reasoning and Professional AI Tools

OpenAI has introduced GPT-6.2 Astra alongside GPT-6.2 Sol, bringing improvements to complex reasoning, coding, scientific research and professional document creation. The new models are available through ChatGPT Work, Codex and the OpenAI API, with lower processing costs for Astra.

Key points

  • GPT-6.2 Astra improves complex reasoning, software development, scientific research and professional document creation.
  • OpenAI claims stronger benchmark performance than GPT-6 Astra at less than one-third of the cost per task.
  • GPT-6.2 Astra and Sol are available through ChatGPT Work, Codex and the OpenAI API for eligible users.

OpenAI has launched GPT-6.2 Astra, its latest advanced artificial intelligence model designed to handle demanding professional tasks, alongside a more efficient model called GPT-6.2 Sol. Announced on 17 November 2026, the release focuses on stronger reasoning, improved coding capabilities, more accurate professional outputs and reduced computational costs.

The company says GPT-6.2 Astra delivers better results across complex workflows while consuming fewer output tokens than its predecessor, GPT-6 Astra. The improvements cover software engineering, scientific research, document creation, computer interaction and advanced design tasks.

GPT-6.2 Astra Brings Improvements to Coding and Professional Work

One of the main improvements in GPT-6.2 Astra is its ability to produce professional documents, spreadsheets and presentations that more closely follow supplied templates and industry conventions.

OpenAI says the model generates clearer written material, more consistent slide layouts and better-structured financial workbooks, reducing the amount of manual editing required.

Software development is another major focus. GPT-6.2 Astra is designed to implement and test complex changes across existing codebases, with improvements in code quality, testing and frontend development.

The model also demonstrates stronger visual reasoning and computer-use capabilities, allowing it to work with engineering design tools, interactive websites and three-dimensional environments.

In OpenAI’s Agents’ Last Exam evaluation, GPT-6.2 Astra scored 5.2 points higher than GPT-6 Astra at less than one-third of the cost per task. The company also reported a nine-percentage-point improvement on Terminal-Bench-Science 0.1, an evaluation of scientific work performed through a terminal.

Agents’ Last Exam evaluates models on long-horizon professional workflows. Its leaderboard results for Fable 5.2 exclude nine tasks that used a fallback model.

These results are based on the evaluation configurations described in OpenAI’s announcement and should not be interpreted as guaranteed improvements across every real-world task.

GPT-6.2 Sol Targets Everyday AI Tasks

Alongside Astra, OpenAI introduced GPT-6.2 Sol as a more economical option for everyday coding, writing and multistep workflows.

Compared with GPT-6.1 Sol, the newer model is designed to complete more work within the same paid usage limits. OpenAI positions it as a practical alternative for tasks that do not require Astra’s highest level of reasoning capability.

The company says Sol also approaches Astra’s performance on certain software engineering evaluations while operating at substantially lower cost.

Pricing and Availability

GPT-6.2 Astra and Sol are available in ChatGPT Work and Codex for eligible Plus, Pro, Enterprise and Edu users. Enterprise and Edu access may require activation by a workspace administrator.

Developers can access the models through the OpenAI API using gpt-6.2-astra and gpt-6.2-sol.

GPT-6.2 Astra costs $8 per million input tokens and $40 per million output tokens, with cached input priced at $0.40 per million tokens. These standard API rates represent a 20% reduction in input and output token prices compared with GPT-6 Astra.

GPT-6.2 Sol costs $2 per million input tokens and $10 per million output tokens, with cached input priced at $0.10 per million tokens.

OpenAI has also introduced expanded Ultrafast access, allowing Pro $200 subscribers to use the faster processing option with Sol, while Pro $500 subscribers can use it with both models.

Stronger AI Capabilities Bring Additional Safeguards

The release also introduces tighter safeguards for advanced cybersecurity and biological applications.

According to OpenAI, both models meet its Critical capability threshold for cybersecurity and High threshold for biological and chemical capabilities. GPT-6.2 Astra is also the company’s first externally released model to meet its High threshold for AI self-improvement.

OpenAI says additional monitoring, restrictions and intervention systems have been introduced to manage risks associated with these capabilities.

The launch extends OpenAI’s focus on AI systems capable of completing complex professional workflows, rather than simply generating individual responses. For developers and organisations, the main practical changes are improved task performance, more efficient token usage and a choice between a higher-capability model and a lower-cost alternative.

Source
KA

Founder and editor of Tech Prolonged since 2010. A software engineer by profession, Khurram covers Google, Android and Microsoft, and edits the newsroom's coverage. Find him on X at @khurramar.