Anthropic has released Claude Opus 4.5, claiming it as the best model in the world for coding, agents, and computer use. The flagship model arrives at a new price point of $5/$25 per million tokens—making Opus-level capabilities more accessible than previous generations.
Benchmark Leadership#
Claude Opus 4.5 leads SWE-bench Verified, the industry-standard evaluation for real-world software engineering tasks. The model outperforms competitors across seven of eight programming languages on SWE-bench Multilingual and shows a 10.6% improvement over Sonnet 4.5 on the Aider Polyglot benchmark.
In a notable internal test, Anthropic reports that Opus 4.5 scored higher than any human candidate on their notoriously difficult performance engineering take-home exam within the prescribed two-hour time limit.
Token Efficiency Gains#
Perhaps the most significant practical improvement is efficiency. At medium effort levels, Opus 4.5 matches Sonnet 4.5’s best SWE-bench score while using 76% fewer output tokens. At maximum effort, it exceeds Sonnet 4.5 by 4.3 percentage points while still using 48% fewer tokens.
Anthropic has introduced a new effort parameter in the API, allowing developers to balance between speed and capability depending on task requirements.
Safety Improvements#
Anthropic claims Opus 4.5 is their most robustly aligned model to date. The model shows significant improvements in defending against prompt injection attacks—a critical concern for agentic workflows where models interact with external systems.
Platform Updates#
Alongside Opus 4.5, Anthropic announced several product enhancements:
- Extended conversations in Claude apps now automatically summarize earlier context
- Claude Code gains improved Plan Mode and desktop app integration for running multiple parallel sessions
- Claude for Excel expands to all Max, Team, and Enterprise users
Early Industry Response#
Partners report substantial improvements. GitHub’s Chief Product Officer noted the model “surpasses internal coding benchmarks while cutting token usage in half.” Cursor’s CEO called it “a notable improvement with improved pricing and intelligence on difficult coding tasks.”
Claude Opus 4.5 is available now via the API using claude-opus-4-5-20251101, as well as through Claude apps and major cloud platforms including Amazon Bedrock and Google Cloud’s Vertex AI.