Anthropic released Claude Opus 4.5 on Monday, completing its three-model family and marking the company’s third major launch in just two months. The new flagship model claims the top spot in coding benchmarks while cutting prices dramatically.
The release caps a rapid-fire rollout that began with Claude Sonnet 4.5 in late September and continued with Claude Haiku 4.5 in October. Now with Opus joining its siblings, Anthropic offers developers a complete toolkit: Opus for complex production work, Sonnet for everyday tasks, and Haiku for speed and efficiency-related tasks that require simple logic.
Claude Opus 4.5 scored 80.9% on SWE-bench Verified, a benchmark testing real-world software engineering tasks. That edges out OpenAI’s GPT-5.1-Codex-Max at 77.9% and Google’s Gemini 3 Pro at 76.2%. Anthropic says Opus outperformed every human candidate on its internal performance engineering exam—a two-hour assessment designed to evaluate judgment under pressure.
There has been a race between AI giants to end the year in the top of the leaderboards. Google launched Gemini 3 Pro on November 18, positioning it as a breakthrough in multimodal reasoning. OpenAI countered the next day with GPT-5.1-Codex-Max.
Anthropic’s response with Opus came just a few days later, but it arrived with a hook: pricing at $5 per million input tokens and $25 per million output tokens, which represents a 67% cut from the previous Opus model.
Alibaba’s Qwen models add another dimension to the race. The company released Qwen2.5-Max in late January with over 20 trillion training tokens, claiming it outperforms DeepSeek-V3 on key benchmarks. Qwen3-Max, launched in September with more than 1 trillion parameters, ranks third globally on LMArena and excels at different tasks like deep research, multimodal reasoning, or workflows in eastern languages. While Qwen models remain relatively obscure in Western markets, they represent China’s push for AI self-reliance amid U.S. chip export restrictions
That pricing sits between the OpenAI’s newest GPT-5.1 ($1.25/$10) and Anthropic’s older Opus 4.1 ($15/$75), though it’s still pricier than Gemini 3 Pro’s $2/$12. The reduction signals market pressure as leading AI labs compete not just on capability, but on making frontier intelligence economically viable for scaled deployment.
Claude’s latest offering is still pricier than many Asian competitors, but is also a bit more capable. So users now have the ability to choose between cost-efficiency or pure technical capability.
Sonnet 4.5, released September 30, brought state-of-the-art coding and agent capabilities at moderate cost and was already better than Opus 4.1 at specific tasks. The simpler Haiku 4.5 was unveiled October 15. Opus 4.5 now sits at the top, handling the hardest reasoning and longest-running tasks.
Similar to Sonnet and GPT-5, Claude Opus 4.5 uses what Anthropic calls a “hybrid reasoning” architecture—a single model trained for both direct inference and chain-of-thought processing. It supports a 200,000 token context window and can output up to 64,000 tokens. The model’s knowledge cutoff is March 2025, slightly ahead of Sonnet’s January date.
Developer Simon Willison tested Opus 4.5 extensively over the weekend, using it to refactor one of his projects. The model handled 20 commits across 39 files, adding 2,022 lines and removing 1,173 others. “It’s clearly an excellent new model,” Willison wrote, though he noted that reverting to Sonnet 4.5 afterward didn’t dramatically reduce his productivity.
“I’m not saying the new model isn’t an improvement on Sonnet 4.5—but I can’t say with confidence that the challenges I posed [to] it were able to identify a meaningful difference in capabilities between the two,” he wrote.
Theo Browne, a developer, YouTuber, and CEO of AI platform T3 Chat called Claude Opus 4.5 “insane,” adding in a video review that it’s “definitely the best coding model ever made.”
The competitive landscape has become increasingly crowded. Google’s Gemini 3 Pro dominated headlines last week, scoring 1501 on LMArena and earning praise from Salesforce CEO Marc Benioff, who said he’s ditching ChatGPT for Google’s model. That announcement sent Alphabet’s stock up more than 6% and reportedly rattled OpenAI CEO Sam Altman, who told colleagues Gemini would create “temporary economic headwinds.”
Microsoft and Nvidia announced multi-billion-dollar investments in Anthropic last week, boosting the startup’s valuation to approximately $350 billion. The deals include expanded Azure integration and Nvidia-powered infrastructure for training and deploying Claude models.
Opus 4.5 is available immediately via Anthropic’s API, AWS Bedrock, Google Vertex AI, and the Claude web and desktop apps.
Generally Intelligent Newsletter
A weekly AI journey narrated by Gen, a generative AI model.
