Anthropic Launches Claude Opus 4.5, Reclaiming the World Best Crown for Coding and Agents

In a massive counterstrike against OpenAI and Google, Anthropic has officially released Claude Opus 4.5, its most powerful artificial intelligence model to date. Unveiled on Monday, November 24, 2025, the new flagship model is being positioned as the undisputed "best model in the world" for software engineering, complex reasoning, and autonomous computer use.

The release comes just weeks after the debut of Google’s Gemini 3 and OpenAI’s GPT-5.1, signaling that the AI arms race is accelerating rather than stabilizing. However, Anthropic’s aggressive pricing strategy—slashing costs by nearly 66%—may be the bigger headline for enterprise customers.

"Smarter Than Our Own Engineers"
The headline achievement for Opus 4.5 is its performance on complex coding tasks. Anthropic claims the model has achieved a score of 80.9% on SWE-bench Verified, a rigorous industry standard for software engineering. This score reportedly beats both Google's Gemini 3 Pro and OpenAI's GPT-5.1.

In a bold blog post, the company revealed that Opus 4.5 also "scored higher than any human candidate ever" on Anthropic’s own notoriously difficult engineering take-home exam.

"Opus 4.5 is a step forward in what AI systems can do," the company stated, emphasizing the model's ability to handle ambiguity, navigate tradeoffs, and execute multi-step plans without the "hand-holding" required by previous generations.

A Massive 66% Price Cut
Perhaps the most significant news for developers is the pricing. Historically, the "Opus" tier has been prohibitively expensive, costing $15 per million input tokens.

With Opus 4.5, Anthropic has aggressively undercut the market. The new pricing is set at $5 per million input tokens and $25 per million output tokens. This represents a roughly 66% price drop compared to the previous Opus 4, making "frontier-level" intelligence accessible for high-volume enterprise workloads for the first time.

Key Features and "Tool Search"
Beyond raw intelligence, Opus 4.5 introduces critical architectural changes designed to solve the "context bloat" that plagues AI agents.

Tool Search Tool: A new feature that allows the model to "discover" the tools it needs on-demand rather than loading every possible tool definition upfront. This can save thousands of tokens per query and makes the model significantly faster and cheaper to run.

Computer Use: Building on the capabilities introduced in Sonnet 4.5, the new Opus model achieves state-of-the-art results on benchmarks like OSWorld, demonstrating an uncanny ability to control desktop computers, navigate websites, and use applications just like a human operator.

200k Context Window: The model retains the massive 200,000-token context window, allowing it to ingest entire codebases or long legal documents in a single prompt.

Availability
Claude Opus 4.5 is available immediately to developers via the Anthropic API and the Claude.ai web interface. In a synchronized rollout, it has also launched simultaneously on all three major cloud platforms: Amazon Bedrock, Google Vertex AI, and Microsoft Foundry.

The release sets the stage for a fierce end-of-year battle for corporate AI budgets, with Anthropic clearly betting that a combination of superior coding skills and drastically lower prices will be enough to lure customers away from the GPT ecosystem.