Anthropic Launches Claude Opus 4.8. Surpasses OpenAI’s GPT 5.5 and Google’s Gemini 3.1 Pro

Tradingkey - On May 28, Anthropic released its latest flagship AI model, Claude Opus 4.8, which is an upgraded version of its flagship product, Opus 4.7.

According to official data from Anthropic, Opus 4.8 outperformed in several AI-centric synthetic benchmarks, surpassing OpenAI's ChatGPT-5.5 and Google's (GOOGL) Gemini 3.1 Pro.

Opus 4.8 Core Technical Upgrade Highlights

In terms of model reliability, Anthropic stated that the pre-Opus 4.8 model is more "honest": it has significantly reduced the probability of presenting false information as fact in the absence of sufficient supporting information. When the model detects "uncertainty in its own work," it proactively informs the user while significantly reducing the occurrence of providing unsubstantiated claims.

In addition to core model upgrades, Anthropic has launched new features across its entire product suite. These include dynamic workflow capabilities, which can simultaneously mobilize hundreds of small sub-agents to collaborate and deliver final results.

Regarding cost and efficiency control, the new features allow users to adjust Claude's "engagement level" when executing tasks, thereby reducing the time or capital spent on specific functions.

For instance, in browsers and Claude Cowork, users can make flexible adjustments to allow the AI to increase its engagement for operations requiring deep thinking, while reducing engagement for tasks that need to be completed quickly. This lower level of engagement will directly reduce the number of tokens consumed by users during AI chats and conversation reviews.

In terms of availability and market pricing, pricing for regular users remains consistent with Opus 4.7, at $5 per million input tokens and $25 per million output tokens. Furthermore, Fast mode is priced at $10 per million input tokens and $50 per million output tokens.