OpenAI launches GPT-6.2 Astra and Sol with performance boost and lower costs
The models add 5.2 percentage points to Astra’s professional‑task score, cut API fees by 20 percent and open a legal‑access mode for enterprise customers.

Key takeaways
- 1Astra scores 5.2 points higher than GPT‑6 Astra on the Agents’ Last Exam
- 2API input price for Astra reduced to $8 per million tokens, a 20 % cut
- 3Sol’s processing cost is roughly one‑fifth of Astra’s per task
- 4Legal‑access mode achieved a 20.8 % score on the LAB‑AA test
OpenAI launched GPT-6.2 Astra and GPT-6.2 Sol on Thursday, November 17, 2026, adding 5.2 percentage points to Astra’s professional‑task benchmark and reducing processing costs by up to 20 percent. The rollout expands the ChatGPT Enterprise suite with two models aimed at different workload types.
Astra outperformed the previous GPT‑6 Astra by 5.2 points on the Agents’ Last Exam, a multistep professional test, and beat Claude Fable 5.2 by 5.5 points. It also scored nine points higher on Terminal‑Bench‑Science 0.1 and set a new high on the Vals Science Index. Sol matched Astra on Cognition’s FrontierCode 2.0 and costs roughly one‑fifth of Astra per task.
OpenAI cut Astra’s API token price by 20 percent, charging $8 per million input tokens and $40 per million output tokens, while cache tokens cost $0.40 each. Sol retained the $2 per million input and $10 per million output rates of its predecessor, making it about one‑seventh the price of Astra for everyday workloads.
The company also introduced tighter safeguards, flagging requests that exceed its critical cybersecurity threshold or high‑risk biological and chemical capabilities. Under the new configuration, every ChatGPT Enterprise customer can request legal‑access mode, which achieved a 20.8 percent score on the Artificial Analysis LAB‑AA test. OpenAI said the safeguards will still allow approved defensive research through its Daybreak programme.
Both models are available to Plus, Pro, Enterprise and Edu users via ChatGPT Work and Codex, and developers can reach them through the OpenAI API with ultrafast processing at launch. The rollout follows OpenAI’s October partnership with Google Cloud, which now sells the models to enterprise customers on the cloud platform.
Lower API fees and expanded legal functionality lower barriers for firms building AI‑driven contracts, compliance tools and research pipelines. The update arrives weeks after OpenAI introduced the ChatGPT Wallet, enabling AI agents to make tokenised payments without exposing card details, a feature that can be combined with the new models for automated legal transactions. Traders and developers will watch OpenAI’s pricing dashboard for further adjustments.
OpenAI’s launch comes as rivals such as SpaceXAI released Grok 4.7 and Anthropic continues to iterate its Claude series. Analysts at TechInsights note that the cost advantage and legal‑access feature could pressure competitors to accelerate similar safeguards. The move also reinforces OpenAI’s position as the leading provider of enterprise‑grade generative AI.
Regulators in the United States and Europe have been scrutinising AI models for misuse. By embedding higher‑threshold safeguards, OpenAI aims to meet emerging guidance from the FTC and the EU’s AI Act, which call for risk‑based controls on high‑impact systems.
Reported and edited by the YourKhabri desk under our editorial policy. Nothing here is investment advice.


