# GPT-6.2 Astra Ranks Third as OpenAI Models Post Lower Task Costs

By Simon Yoon

Canonical URL: https://www.tokenpost.com/news/technology/60715
Published: 2026-11-18T18:06:47.000Z
Updated: 2026-11-18T18:06:47.000Z
Section: Technology

> A Nov. 18 snapshot put Astra at 1,781 points and Sol at 1,757. Astra’s median task cost was about 3.8 times Sol’s.

GPT-6.2 Astra (Max) scored 1,781 points, placing third, while GPT-6.2 Sol (Max) earned 1,757 for seventh. In WebDev standings dated Nov. 18, 2026, both models outscored earlier versions at a lower median cost per task.

Astra gained 18 points over GPT-6 Astra (Max) as its median task cost fell 42%, from $9.46 to $5.45. Sol beat GPT-6.1 Sol (Max) by 13 points; its median task cost declined 21%, from $1.84 to $1.45.

Astra led Sol by 24 points but had about 3.8 times its median task cost. Both models sat on the leaderboard’s cost-performance boundary: no competing model paired a higher score with a per-task cost no greater than theirs.

The standings draw on community evaluations. Agent scores remained pending, leaving performance on multi-step AI tasks outside this snapshot.

OpenAI launched the models Nov. 17. Astra’s standard input and output API rates were each 20% below GPT-6 Astra’s, with cached input 60% lower; details are in [Astra’s API pricing update](<https://www.tokenpost.com/news/technology/59729>).



## Links in this article

- [Astra’s API pricing update](https://www.tokenpost.com/news/technology/59729)
