Astra & Sol debut w/ stronger coding + lower cost. Waiting for specific SWE-bench numbers
OpenAI releases GPT-6.2 Astra and Sol with improved coding and efficiency
Original: Introducing GPT-6.2 Astra and Sol
Importance: 新世代モデルの主要リリースで、コーディング性能向上と低コスト化により多くのAPI・ChatGPT Workユーザーに直接影響
Summary
OpenAI introduces GPT-6.2 Astra and Sol, featuring enhanced professional outputs, coding capabilities, computer use, and efficiency at lower cost across ChatGPT Work, Codex, and the API. This marks a significant performance upgrade with improved cost-effectiveness.
Key Points
- Astra and Sol dual model lines released
- Enhanced coding performance improvements
- Computer use capability included
- Inference cost reduction achieved
- Available on ChatGPT Work, Codex, and API immediately
View developer notes (APIs, breaking changes, migration)
GPT-6.2 splits into Astra (high-performance general) and Sol (efficiency variant). Emphasizes enhanced coding performance, computer use capabilities, and inference efficiency. Deployed across ChatGPT Work, Codex, and standard API endpoints. Achieves lower cost-per-inference vs. prior generation while improving benchmark scores. Specific context window, token pricing, and breaking API changes not detailed in announcement. Developer migration expected to be drop-in compatible.
Source: https://openai.com/index/introducing-gpt-6-2-astra-sol
Outlet: OpenAI News
This article is an AI-generated summary (OpenAI GPT-4o-mini) of publicly available information from Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek, Sakana, and other vendors. The original source URL is always provided in accordance with fair-use citation requirements. Summaries are AI-generated and may contain mistranslations or misinterpretations. Always verify details with the original source.