Anthropic launches Claude Opus 4.8, beats GPT-5.5 and Gemini 3.1 Pro
Outperforms rivals in agentic coding, finance, and computer use benchmarks.
Anthropic has introduced Claude Opus 4.8, its latest flagship AI model, as competition with OpenAI and Google intensifies ahead of potential public listings. According to the company, Opus 4.8 outperforms OpenAI’s GPT-5.5 and Google’s Gemini 3.1 Pro across synthetic benchmarks in agentic coding, financial analysis, and computer use—categories critical for enterprise automation. Anthropic emphasizes that its model is more 'honest,' meaning it is less likely to present unsupported information as fact and will communicate uncertainties when data is lacking. This design aims to reduce hallucinations and increase trustworthiness in responses.
Alongside Opus 4.8, Anthropic launched dynamic workflow features that can deploy hundreds of smaller subagents simultaneously to complete complex tasks. Users can now control how much effort Claude applies to a task—either spending more resources for thoroughness or using a faster, token-saving mode. This token control allows businesses to manage costs more effectively. Pricing stays consistent with Opus 4.7: $5 per million input tokens and $25 per million output tokens for regular use, with Fast mode at $10 input and $50 output per million tokens. The model is available immediately across Anthropic’s platforms.
- Claude Opus 4.8 outperforms GPT-5.5 and Gemini 3.1 Pro on agentic coding, finance, and computer use benchmarks.
- New dynamic workflow feature can deploy hundreds of subagents simultaneously for complex task automation.
- Pricing remains $5/$25 per million tokens (regular) and $10/$50 (fast mode), with user-controlled effort levels to reduce token consumption.
Why It Matters
Enterprises gain a more honest, controllable AI model that excels at autonomous task execution, lowering costs and risk.