Anthropic and OpenAI Drop New High-Efficiency Models
The new models prioritize faster speeds and lower prices.
Two leading artificial intelligence companies, OpenAI and Anthropic, unveiled new high-efficiency models on Tuesday, both promising increased speed and reduced costs compared to their predecessors. Anthropic introduced Opus 5.5, which matches the capabilities of Fable 5.1 but costs nearly 40% less than Opus 5. OpenAI also rolled out GPT-6 Sol and GPT-6 Luna, follow-ups to the July-released 5.6 versions, which aim to make the power of ChatGPT Astra more accessible at lower budgets while offering higher usage limits for business and developer workflows.
OpenAI's new ChatGPT models emphasize factuality, providing more accurate answers following user prompts, with a reduced mistake rate of over half compared to previous models. Despite OpenAI's benchmarks suggesting efficiency gains, the impact for everyday users may be less apparent. Both GPT-6 Sol and Luna will be deployed on ChatGPT Work and Codex, but free users and lower-tier subscribers will still have access to GPT-6 Luna.
Anthropic's Claude Opus 5.5 model is positioned as its strongest-performing in terms of alignment, meaning it better adheres to human intentions, values, and rules. This does not guarantee perfection, and its agentic capabilities will be further tested. Anthropic maintains the same safeguards across its models, which may revert to older versions for specific cybersecurity or scientific tasks.
The new models from both companies require less compute power than their predecessors, leading to cost savings of up to 40% for typical workloads. Anthropic has also increased usage limits on Pro, Max, and Team plans to 5 hours for default settings. Additionally, both Anthropic and OpenAI claim their models communicate more naturally, improvements noted through user feedback.
Written by urgent.news from CNET's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
