Anthropic Launches Claude Opus 4.8: Revolutionary AI Enhancements in Coding and Reasoning

Liputan6.com, Jakarta - Anthropic officially launched Claude Opus 4.8 on May 28, 2026.

This latest artificial intelligence (AI) model is a significant upgrade from the previous version, Claude Opus 4.7, promising more advanced and efficient capabilities.

This launch marks a major step forward for Anthropic in providing more powerful AI tools for a variety of needs.

This upgrade aims to provide users with more effective AI collaborators, with a focus on coding capabilities, agent skills, and reasoning.

Claude Opus 4.8 comes with a variety of new features and performance optimizations designed for complex tasks.

This model builds on Opus 4.7 with improvements across various benchmarks.

This model is already widely accessible through claude.ai, the Claude API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.

Before we ship a new model, these teams try to break it.They build with it, push it to its limits, and tell us where it falls short. What they find makes the final model better. pic.twitter.com/Q4LD01zIIc

— Claude (@claudeai) May 28, 2026

Claude Opus 4.8 represents a significant leap in coding capabilities.

This model achieved a score of 69.2% on SWE-bench Pro, an increase from 64.3% on Opus 4.7.

This capability allows Claude Opus 4.8 to produce production-ready code with minimal supervision, with careful planning, and to operate reliably on larger codebases.

Furthermore, the agent's performance in Opus 4.8 has also significantly improved.

On the Super-Agent benchmark, Claude Opus 4.8 was the only model to solve every case thoroughly, even outperforming the previous Opus model and GPT-5.5 at comparable cost.

Improvements are seen in the knowledge work benchmark, from 1753 to 1890, and in financial analysis, from 51.5% to 53.9%.

Anthropic emphasizes that Opus 4.8 is more "honest," significantly reducing the likelihood of presenting false information as fact.

The model even proactively notifies users when it detects "uncertainty in its work."

Claude Opus 4.8 introduces several innovative features to improve the user experience.

One of these is "Effort Control," which allows users in claude.ai to adjust Claude's effort level in completing tasks.

This option provides flexibility between faster/cheaper or deeper/slower responses.

The "Dynamic Workflows" feature in Claude Code enables handling very large-scale problems by scheduling jobs and running hundreds of parallel sub-agents in a single session, although this feature is still in research preview.

Additionally, the Messaging API now supports system messages mid-conversation, allowing developers to update instructions without corrupting the prompt cache.

This model also features a "Fast Mode," which is 2.5 times faster and three times cheaper than the previous model's fast mode.

Fast mode is priced at $10 per million input tokens and $50 per million output tokens, while the regular price remains at $5 per million input tokens and $25 per million output tokens.

Claude Opus 4.8 supports a context window of 1 million tokens by default on the Claude API, Amazon Bedrock, and Vertex AI, with a maximum output window of 128,000 tokens.

Official Anthropic data shows that Opus 4.8 outperforms OpenAI's ChatGPT-5.5 and Google's Gemini 3.1 Pro in several synthetic AI benchmarks.

This model is also the strongest computer-based and browser-agent model tested, scoring 84% on Online-Mind2Web, a significant improvement over Opus 4.7 and GPT-5.5.

In some benchmarks, Claude Opus 4.8 even leads the Artificial Intelligence Analysis Index.

While Opus 4.8 is a very capable model, Anthropic is not stopping there.

They are working on a Mythos-class model that promises even higher intelligence.

The Mythos Preview model is currently used by several organizations for cybersecurity work as part of Project Glasswing.