Major artificial intelligence laboratories keep releasing new models quickly despite recent industry calls to slow down development. An open letter from AI company employees in July asked frontier labs to slow their pace. Anthropic CEO Dario Amodei echoed this message earlier this month.
"We must slow the pace at which we improve the capabilities of AI models," Amodei said in an online post on September 12, 2026. He reiterated the message from AI staff who signed an open letter titled Pacing the Frontier. Despite these warnings, major model releases continue to appear frequently.
Anthropic and OpenAI each released new versions of their Claude and GPT models on Tuesday. Anthropic launched Opus 5.5, calling it the strongest-performing model tested to date. The release arrived just 21 days after Fable 5.1 and Mythos 5.1. Those models followed Opus 5 by 39 days. Anthropic's release cadence has accelerated from roughly every quarter in 2025 to almost monthly in 2026.
OpenAI slowed its release cycle somewhat after a flurry of updates early this year. The company delivered several model variants in February and March, followed by GPT-5.5 in April. GPT-5.6 reached general availability 77 days later on July 9. GPT-6 Astra debuted on September 3, 2026, followed by GPT-6 Sol and Luna 19 days later.
The competitive releases arrive as both rival companies prepare for initial public offerings. Anthropic could go public before the end of the year. OpenAI is not expected to launch an initial public offering until 2027.
Anthropic states that Opus 5.5 performs better than its predecessor on software tasks. The model allows speedy code migrations and can reduce software load times. Artificial Analysis puts Opus 5.5 on par with GPT-6 Astra for agentic work. The Claude model claimed the top spot in the Artificial Analysis Intelligence Index with a score of 58 across ten benchmark evaluations.
Anthropic notes that Opus 5.5 is comparable to Mythos 5.1 in cybersecurity and biology capabilities. Because of this, the system can fall back to earlier models when safety safeguards trigger. The new Claude model is also cheaper than its predecessor.
"Our tests show that at default settings it will cost 40 percent less than Opus 5 on typical workloads," Anthropic stated in its announcement. Input and output tokens cost $4 and $20 per million, representing a 20 percent reduction from Opus 5. Cache reads cost $0.20 per million tokens, down 60 percent. Opus 5.5 also generates output more than 30 percent faster than Opus 5.
OpenAI addressed costs with GPT-6 Sol and Luna by leaving high-performance workloads to Astra. Improvements in caching and inference let the company serve these models at lower cost. OpenAI is passing those savings directly to users by reducing API prices for Sol and Luna by 50 percent compared with their GPT-5.6 promotional pricing.
Artificial Analysis found that GPT-6 Sol costs $1.06 per task compared to $1.99 per task for the older GPT-5.6 Sol. GPT-6 Luna costs $0.07 per task compared to $0.18 for GPT-5.6 Luna. OpenAI believes better pricing will win business from its rival.
Both companies claim their models improve alignment and safety. Anthropic's defenses also aim to protect its business from copycats. Opus 5.5 includes preserved thinking, a defense against model distillation introduced with Fable 5.1. This feature prevents API customers from editing prior context to obtain the model's chain of reasoning.
Initial public offerings for both Anthropic and OpenAI remain the next known business milestones as the companies continue their accelerated release schedules.
