Anthropic and OpenAI Simultaneously Launch New AI Models with Lower Costs
Anthropic and OpenAI have simultaneously launched advanced AI models, introducing Opus 5.5 alongside GPT-6 Sol and Luna to boost performance and cut costs by up to 50%.
Anthropic and OpenAI have simultaneously launched new artificial intelligence models aimed at two primary goals: improving performance and significantly reducing usage costs. The timing between the two companies was notably tight, with Anthropic unveiling Opus 5.5—the successor to Fable 5.1—just about 90 minutes before OpenAI announced the expansion of its GPT-6 family.
OpenAI Expands GPT-6 Lineup with Sol and Luna
OpenAI introduced updated versions of its two smaller models, Sol and Luna, joining GPT-6 Astra, which was announced earlier this month as the company's most powerful model. According to the company, the new Series 6 models are offered via API at prices up to 50% lower than the comparable Series 5.6.
"Please welcome GPT-6 Sol and Luna to the GPT-6 universe. They build on the advances behind GPT-6 Astra, bringing much of its strengths into faster and more affordable models to support work at scale," OpenAI stated.
The cost reduction is primarily attributed to improvements in inference mechanisms and prompt caching, allowing for faster responses and the reuse of existing data. Sol is designed for complex tasks, primarily coding, while Luna is tailored for routine office work at scale, such as document summarization, data extraction, and rapid response.
Anthropic Counters with Opus 5.5
On the other side, Anthropic has tailored Opus 5.5 primarily for enterprise clients requiring complex workflows, including identifying and repairing code inefficiencies, conducting financial analyses, and solving business problems. According to Anthropic's benchmark data, Opus 5.5 outperformed GPT-6 Astra in agentic coding across two dedicated tests: Terminal-Bench 4.0 and FrontierCode v1.1.
"Introducing Claude Opus 5.5, the first model in our new Claude 5.5 family. It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.5," Anthropic posted on social media.
Both companies also emphasized advancements in their models' protection and reliability mechanisms. Anthropic noted that Opus 5.5 writes more clearly, and user jailbreak attempts have dropped by about 85% compared to previous generations. Meanwhile, OpenAI reported that its new models are better aligned with safety tests, exhibiting fewer false positives on code execution and more frequently refusing unsafe prompts.
