Breaking
OpenAI Launches GPT-6.2 Astra and Sol With 20% Lower API Pricing
OpenAI released GPT-6.2 Astra and GPT-6.2 Sol on 17 November 2026, claiming 5.2-point benchmark gains at less than one-third the cost per task of the previous Astra model.

By TechWhoKnows
3 min read
OpenAI launched GPT-6.2 Astra and GPT-6.2 Sol on 17 November 2026, introducing upgraded models for complex professional work, coding, scientific research and automated computer tasks.
The company said GPT-6.2 Astra delivers stronger performance than its predecessor while reducing API input and output token prices by 20%, from $10 to $8 per million input tokens and from $50 to $40 per million output tokens. Cached input costs $0.40 per million tokens, a 60% reduction.
On Agents' Last Exam, a benchmark evaluating complex professional workflows, GPT-6.2 Astra scored 5.2 percentage points higher than the original GPT-6 Astra at less than one-third of the cost per task. OpenAI said the model also outperformed Anthropic's Claude Fable 5.2 by 5.5 points at approximately 5% of the cost, according to OpenAI's evaluations.
The model also improved performance on Terminal Bench Science 0.1 by nine percentage points over GPT-6 Astra, while setting new highs on benchmarks measuring professional document creation, scientific research and interactive 3D design. OpenAI highlighted Astra's ability to build financial valuation workbooks, design websites, navigate engineering software and carry out complex workflows with fewer revisions.
Alongside Astra, OpenAI introduced GPT-6.2 Sol, a more affordable model optimised for everyday tasks and coding. The company said Sol approaches Astra's performance on the FrontierCode 2.0 benchmark at approximately one-fifth of the cost. GPT-6.2 Sol retains API pricing of $2 per million input tokens and $10 per million output tokens.
Both models introduce stronger cybersecurity capabilities, prompting OpenAI to implement tighter safeguards around potentially dangerous requests. The company said GPT-6.2 Astra is its first externally released model to meet the High threshold for AI self-improvement under its Preparedness Framework.
The models are rolling out through ChatGPT Work and Codex for eligible Plus, Pro, Enterprise and Edu users, with API access available to developers under the model names gpt-6.2-astra and gpt-6.2-sol.
OpenAI originally introduced GPT-6 Astra in September, positioning it as its most capable model for computer use, coding, cybersecurity and scientific research. The latest release builds on those capabilities while focusing on more efficient execution and lower operating costs.
Sources
- 1CryptoBriefingcryptobriefing.com
- 2Shattered.io analysisshattered.io