Latest Model Information
This article covers the Opus 4.1 release. As of February 2026, the even more powerful Claude Opus 4.6 has been released. For the latest benchmark comparisons, see LLM Coding Benchmark Comparison 2026 Edition.
Claude Opus 4.1 Release - Achieving 74.5% Coding Performance, Immediate Migration from Opus 4.0¶
📢 Overview¶
On August 5, 2025, Anthropic officially released Claude Opus 4.1. This latest flagship model, as a direct successor to Opus 4.0, achieved cutting-edge performance of 74.5% in coding tasks and shows significant improvements in real-world software engineering tasks.
🚀 Major Performance Improvements¶
Dramatic Coding Performance Enhancement¶
| Model | SWE-bench Verified | Agentic Terminal Coding | AIME 2025 |
|---|---|---|---|
| Opus 4.1 | 74.5% | 43.3% | 78% |
| Opus 4.0 | 72.5% | 39.2% | 75.5% |
| Sonnet 3.7 | 62.3% | - | - |
- SWE-bench Verified: Achieved 74.5% on real-world software engineering tasks (industry-leading performance)
- Agentic Terminal Coding: Improved to 43.3% (4.1 percentage point improvement from Opus 4.0's 39.2%)
- Reasoning Ability (AIME 2025): Recorded 78% (2.5 percentage point improvement from Opus 4.0's 75.5%)
Notable Improvements¶
- Significant performance enhancement in multi-file code refactoring
- Hybrid reasoning system: Operates regardless of extended thinking capabilities
- Maintains 200K token context window
- Supports up to 32K token output
💰 Pricing and Usage Limits¶
Pricing Structure (No Changes)¶
Claude Opus 4.1 is offered at the same pricing as Opus 4.0:
- Pro Plan: $20/month
- Max Plan: $100-200/month
- Team Plan: Enterprise pricing
- Enterprise Plan: Custom pricing
Detailed Usage Limits¶
Important: Usage limits are completely identical to Opus 4.0.
Pro Plan ($20/month)¶
- May reach limits at approximately 2 messages per 4 hours
- Unsuitable for large codebases or complex tasks
Max Plan ($100-200/month)¶
- 5x Plan: Minimum 225 messages per 5 hours
- 20x Plan: Minimum 900 messages per 5 hours
- May reach limits in 2-3 hours when running multiple agents
- Normal usage allows 5-6+ hours of operation
🔧 Available Platforms¶
Claude Opus 4.1 is available on the following platforms:
- Claude.ai: For Pro, Max, Team, and Enterprise users
- Anthropic API: Developer API access
- Claude Code: Integrated development environment usage
- Amazon Bedrock: AWS cloud environment
- Google Cloud Vertex AI: GCP cloud environment
- GitHub Copilot: Available as public preview
⚠️ Important Migration Considerations¶
Opus 4.0 Deprecation Schedule¶
- Deprecation Date: August 20, 2025 (15 days after announcement)
- Recommendation: Urgent migration to Opus 4.1 recommended
- Compatibility: Opus 4.1 designed as a drop-in replacement for Opus 4.0
Migration Benefits¶
- Performance Improvement: Enhanced across all benchmarks
- Price Unchanged: No additional costs
- Full Compatibility: No changes to APIs or interfaces
🔮 Future Outlook¶
Anthropic announced on official social media that they "plan to release significant model improvements within the next few weeks". This suggests that Opus 4.1 is an incremental improvement with larger updates forthcoming.
📊 Competitive Comparison¶
Claude Opus 4.1's SWE-bench Verified score of 74.5% represents industry-leading performance. This means it can complete coding tasks with higher accuracy than any other model in real-world software engineering scenarios.
Summary¶
Claude Opus 4.1 represents a natural evolution from Opus 4.0, showing particularly notable improvements in coding capabilities. With no changes to pricing or usage limits, existing Opus 4.0 users can migrate immediately without risk.
However, the strictness of Pro Plan usage limits remains a challenge, and Max Plan or higher is practically necessary for serious development work.
Last updated: August 6, 2025Sources: Anthropic official announcements, GitHub Changelog, various tech media