Anthropic and OpenAI keep shipping models despite calls to slow down

Anthropic and OpenAI each shipped a new flagship model on Tuesday: Claude Opus 5.5 and GPT-6 Sol and Luna. The releases arrived days after Anthropic CEO Dario Amodei publicly urged the industry to slow the pace of capability development.

Slowdown calls, quick releases

In a post published on September 12, 2026, Amodei wrote that "we must slow the pace at which we improve the capabilities of AI models," echoing a July open letter from employees at AI companies who signed it under the name "Pacing the Frontier." Ten days after that post, Anthropic introduced Opus 5.5 and called it "the strongest-performing model we've tested to date."

An accelerating release cadence

Opus 5.5 landed just 21 days after Fable 5.1 and Mythos 5.1, which in turn followed Opus 5 by 39 days. The cadence of Anthropic releases has accelerated from roughly quarterly in 2025 to almost monthly in 2026. OpenAI slowed somewhat after a flurry of model variants in February and March, then shipped GPT-5.5 in April. GPT-5.6 reached general availability 77 days later, on July 9, and GPT-6 Astra debuted on September 3, followed by GPT-6 Sol and Luna 19 days after that. The rival releases come as both companies prepare for initial public offerings: Anthropic's could happen before the end of the year, while OpenAI is not expected to go public until 2027.

Benchmarks, prices, safeguards

Anthropic says Opus 5.5 outperforms Opus 5 on software tasks, enabling faster code migrations and shorter software load times. Artificial Analysis puts the model on par with GPT-6 Astra for agentic work and ranks it first on its Intelligence Index with a score of 58 across 10 benchmark evaluations. At default settings it costs 40 percent less than Opus 5 on typical workloads; input and output tokens are priced at $4 and $20 per million, 20 percent below Opus 5, while cache reads fall to $0.20 per million, 60 percent cheaper. Anthropic also says the model generates output more than 30 percent faster.

OpenAI cut API prices for Sol and Luna by 50 percent against GPT-5.6 promotional pricing, leaving high-performance workloads to Astra. Artificial Analysis measured GPT-6 Sol (max) at $1.06 per task, down from $1.99 for GPT-5.6 Sol, and GPT-6 Luna (max) at $0.07, down from $0.18. OpenAI says that on AutomationBench, a test of business workflows across apps, GPT-6 Sol at xhigh effort outperforms Claude Opus 5 at max effort at just 9 percent of Opus 5's cost per task.

On safety, Anthropic says Opus 5.5 is comparable to Mythos 5.1 in cybersecurity and biology, so it can fall back to earlier models when safeguards trigger. It also ships "preserved thinking," a defense against model distillation introduced with Fable 5.1 that stops API customers from editing Claude's prior context to obtain the model's chain of reasoning. The protection doubles as a guard for Anthropic's business against copycats.