OpenAI fires back at Claude with GPT-6.2 Astra at $8 per million tokens

OpenAI's GPT-6.2 Astra costs $8 per million input tokens, $0.40 cached, $40 output — undercutting GPT-6 Astra while matching Claude Fable 5.2 a week after Anthropic's launch.

Wire notes

  • GPT-6.2 Astra costs $8 per million input tokens, $0.40 cached input, $40 output.
  • Sol is priced at $2 input, $0.10 cached, $10 output per million tokens, matching its predecessor.
  • The launch comes one week after Anthropic released Claude Fable and Mythos 5.2.
  • Anthropic claims Fable 5.2 beats GPT-6 Astra, Opus 5.5 and Gemini 4 Carbon across benchmarks.
  • OpenAI says ChatGPT has hit roughly 20% of Google Search volume.
OpenAI releases GPT-6.2 Astra to regain performance lead from Claude Fable 5.2
PhotoOpenAI releases GPT-6.2 Astra to regain performance lead from Claude Fable 5.2 — frabre / pixabay

OpenAI has released GPT-6.2 Astra and Sol, pricing Astra at $8 per million input tokens, $0.40 for cached input and $40 for output — significantly cheaper than its predecessor, GPT-6 Astra. The launch lands exactly one week after Anthropic rolled out Claude Fable and Mythos 5.2, and days after Google announced Gemini 4 Hydrogen. The frontier model race, in other words, has compressed to a weekly cadence.

The pitch is straightforward: match Anthropic's newest models on benchmarks while undercutting them on cost. OpenAI's own positioning says the new pair delivers performance "more or less" on par with Claude's latest — at what it calls a drastically lower price point.

What do the two models do?

OpenAI splits the release into two tiers. Sol is the volume player. The company describes it as "smarter, more efficient, and lets you complete more tasks on the same paid plan," and positions it as the "go-to model for everyday work."

Astra is the premium tier, tuned for professional output. OpenAI's claim, verbatim:

GPT‑6.2 Astra sets a new quality bar for AI in complex professional work. Its slides, spreadsheets, and documents read less like AI, and more closely follow the templates you provide and the conventions of your field. It writes cleaner code and builds frontends that look and work better, particularly for websites, gaming, and 3D design.

In plain terms: Sol handles the daily grind, Astra handles the client-facing work.

How did Anthropic set the bar?

The competitive pressure is real. Anthropic launched Claude Fable and Mythos 5.2 last week, calling them "designed to tackle the most daunting technical challenges" with improvements in coding, reasoning, and vision.

Anthropic has also published benchmark claims that sting: Fable 5.2 beats Opus 5.5, GPT-6 Astra, and Gemini 4 Carbon "virtually across the board" across coding, cybersecurity and knowledge work, according to the company's own numbers.

That's the claim OpenAI is answering. Astra and Sol are now available to all Plus, Pro, Enterprise, and Edu users inside ChatGPT Work and Codex.

What does it cost?

Per million tokens:

  • GPT-6.2 Astra: $8 input, $0.40 cached input, $40 output
  • Sol: $2 input, $0.10 cached input, $10 output

Sol matches its predecessor's pricing exactly. Astra comes in "significantly more affordable" than GPT-6 Astra, per OpenAI — which matters for enterprise buyers running document and code generation at scale, where output tokens dominate the bill.

The takeaway: near-frontier performance at half the flagship price is now OpenAI's core argument, and it's aimed straight at procurement teams comparing Claude invoices.

Why the timing matters for marketers

OpenAI separately claims ChatGPT has now reached roughly 20% of Google Search volume — a number that puts AI assistants squarely inside the media-planning conversation, not on its margins.

There's movement on distribution too. OpenAI's GPT models are coming to Google Antigravity "soon" and to Google Cloud "in the coming weeks," meaning the same model family will soon sit inside a rival's own developer ecosystem.

For brands, the practical shift is cheaper professional-grade generation: pitch decks, landing pages, 3D assets and frontends built on Astra's pricing floor rather than a premium one.

Watch the next benchmark round — if Anthropic answers within days again, the price war, not the leaderboard, becomes the story.

via 9to5google.com (Source)

Filed under

Share this article:

More from James Calloway

James Calloway

Show full bio

Market editor covering technology and product news in marketing, advertising, marketing technology, media, SEO and search at Marketing Herald.

355 articles

More on the wire

« Previous articleNext article »