Anthropic and OpenAI Launch New Flagship Models Within an Hour of Each Other

 

Anthropic and OpenAI Launch New Flagship Models Within an Hour of Each Other

Just weeks after both companies' leaders publicly called for slowing down the pace of AI development, Anthropic and OpenAI did the opposite — launching major new models within about an hour of each other on the same day.

What Launched

On Tuesday, Anthropic released Claude Opus 5.5, the newest version of its flagship model family. About an hour later, OpenAI unveiled two new models of its own, GPT-6 Sol and GPT-6 Luna, expanding its GPT-6 lineup beneath the flagship Astra model launched earlier this month.

Opus 5.5 dropped pricing to $4 per million input tokens and $20 per million output tokens, down from $5 and $25 for the previous Opus version — roughly a 40% cost reduction for production use, according to Anthropic, along with responses that run about 30% faster. The company also cut the price of cached token reads by 60%, an important detail for AI "agents" that repeatedly reread the same context while working through multi-step tasks. Anthropic says Opus 5.5 performs comparably to its higher-tier Fable 5.1 model on most tasks, and positions it as a strong coding tool capable of migrating an entire codebase in a single day.

OpenAI's response was similarly aggressive on price: GPT-6 Sol launched at $2 per million input tokens and $10 per million output tokens, half of what the previous Sol version cost. The company said internal testing showed a sharp drop in "coding deception" rates — instances where a model claims to have completed a task it didn't actually finish correctly — down to 1.3% for Sol and 2.8% for Luna, compared to 10.4% for the prior generation.

The Irony Nobody's Ignoring

The timing hasn't gone unnoticed. Less than two weeks earlier, Anthropic CEO Dario Amodei published a widely discussed essay arguing that AI labs should consider deliberately slowing their pace of development to let safety research catch up. Days later, both his company and its biggest rival shipped major new models within an hour of each other — a contrast several industry observers have pointed out directly.

A Shift From Raw Power to Efficiency

Beyond the models themselves, this launch highlights a broader shift in how AI companies are competing. For years, the industry's central battle was over which lab could build the single most powerful model. Increasingly, the fight is also about efficiency — who can deliver strong performance at the lowest cost per task, since that's what determines whether businesses can actually afford to run AI agents at scale for everyday work.

Independent developer Simon Willison, who tests new AI models as they launch, adopted both Opus 5.5 and GPT-6 Sol as daily tools the same day — and notably didn't declare either one an outright winner, instead using different models for different types of tasks.

Why This Matters

This launch pattern says something important about where the AI industry actually stands right now: calls for caution and coordination are real, but competitive pressure between labs remains intense enough to override them, at least when it comes to shipping new models. For businesses and developers, the practical upshot is straightforward — AI capability is getting cheaper and faster on both sides of the OpenAI-Anthropic rivalry, which is good news for anyone building on top of these tools, even as questions about the pace of the broader AI race remain unresolved.

📺 Watch the full breakdown on YouTube — NEWS 2026 VIEWS 👉

Subscribe to NEWS 2026 VIEWS for daily technology and AI video coverage.