Despite calls from AI employees and Anthropic's own CEO Dario Amodei for frontier AI labs to slow development, Anthropic and OpenAI both released new models on the same day, contradicting stated concerns about the pace of advancement. Amodei had reiterated the message from the July "Pacing the Frontier" open letter on 12 September, but Anthropic unveiled Opus 5.5 just ten days later, whilst OpenAI released GPT-6 Sol and Luna on the same Tuesday. This underscores the competitive pressure driving both companies to accelerate model development as they prepare for initial public offerings.

Anthropic's release cadence has accelerated dramatically from roughly quarterly in 2025 to almost monthly in 2026, with Opus 5.5 arriving 21 days after its predecessor models. The new models offer significant improvements and cost reductions: Opus 5.5 is 40 per cent cheaper overall, with cache reads 60 per cent cheaper and outputs 30 per cent faster than Opus 5, whilst ranking top on the Artificial Analysis Intelligence Index. OpenAI reduced API prices for GPT-6 Sol and Luna by 50 per cent compared to GPT-5.6, with GPT-6 Sol costing $1.06 per task versus $1.99 previously. Both companies emphasise safety improvements, though Anthropic's "preserved thinking" feature also defends against model copying through interrogation.