OpenAI launches two GPT‑6.2 models with fourfold price gap
OpenAI launches two GPT‑6.2 models with fourfold price gap
OpenAI released two GPT‑6.2 variants, Astra and Sol, with a fourfold price difference between their API rates this week.
Models and pricing
Astra is positioned for complex tasks such as code generation, presentations and web design, while Sol targets routine everyday workflows.
In the API, generating content costs $40 per 1 M tokens for Astra and $10 per 1 M tokens for Sol, reflecting a clear pricing tier.
OpenAI set Astra's price 20% lower than the previous GPT‑6 Astra release to improve cost efficiency for demanding workloads without altering the model family identifier.
Performance and availability
On the Pro tier, the Ultrafast mode accelerates generation up to 4.5×, and correspondingly consumes the usage quota more quickly than standard modes.
Both GPT‑6.2 variants are available via the API and are rolling out to ChatGPT Work and Codex across the Plus, Pro, Enterprise and Edu plans on a phased basis.
Deployment notes
The release maintains separate offering profiles for large-scale and everyday users, enabling customers to choose models that balance performance and cost according to workload demands.
- Astra: optimized for high-complexity tasks and higher throughput requirements.
- Sol: optimized for general-purpose, cost-sensitive use cases.
OpenAI's update clarifies pricing tiers for GPT‑6.2 and brings a lower-cost option for heavy usage while preserving a premium variant for intensive applications.
Related posts

