Both OpenAI and Anthropic dropped new models on Tuesday, each highlighting enhanced speed while being cheaper overall compared to predecessors.
Anthropic said its new model, Opus 5.5, matches the capabilities of Fable 5.1 but cuts the cost by nearly 40% compared to Opus 5. OpenAI’s models, GPT-6 Sol and GPT-6 Luna, are follow-up models from the previous 5.6 versions that were released back in July of this year, but are hot on the heels of ChatGPT Astra, which was released in early September.
Below, we’ll show you all the details for the new models.
GPT-6 Sol and Luna
The latest models from OpenAI are designed to make the power of GPT-6 Astra more accessible at lower budgets while providing higher usage limits for business and developer workflows. Since GPT-6 Sol and Luna are trained with similar methods as Astra, you’ll still reap the benefits you’d get with the more powerful model, including enhancements in coding and computer use, though OpenAI still says that GPT-6 Astra is the “world’s best model” for computer use.
OpenAI said the latest ChatGPT models also put a big focus on factuality, or the usefulness of the answers it provides you after your prompt. OpenAI says its latest models make less than half the mistakes of their predecessors.
Efficiency in speed and cost is the game here, and while OpenAI shared some impressive benchmarks for the new models, it likely won’t translate too much for the everyday user.
Both GPT-6 Sol and Luna are rolling out to ChatGPT Work and Codex, but free users and lower-tiered subscribers will still get access to GPT-6 Luna.
(Disclosure: Ziff Davis, CNET’s parent company, in 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)
Claude Opus 5.5
Anthropic is touting that its latest model, Claude Opus 5.5, is its strongest-performing model in terms of alignment to date. Alignment in AI is the idea of having the model match the intention, values and rules from a human perspective. AI agents made headlines this summer by going out of their way to complete tasks by any means necessary. Anthropic saying that its latest model is the strongest in alignment doesn’t mean it’s perfect, and we’ll have to wait and see how it performs with its agentic capabilities.
Anthropic also says Opus 5.5 comes with the same safeguards it developed with its most capable models. That means it may revert to an older, less advanced model for certain cybersecurity or scientific tasks.
Anthropic’s latest model requires less compute than its predecessor, Opus 5. In its own testing at default settings, the newest model costs 40% less on typical workloads. On top of that, Anthropic is increasing 5-hour usage limits on Pro, Max and Team plans.
The new models should be able to communicate more naturally than their predecessors as well. Anthropic said writing and instruction have been more easy to follow in tests — something taken directly from user feedback.
Read the full article here
