Anthropic Launches Claude Opus 5, Promising Advanced Capabilities at Lower Cost
Anthropic has introduced its latest large language model, Claude Opus 5, which is described as a "thoughtful and proactive model" that approaches the intelligence of Claude Fable 5 but at half the price. The new model is currently topping the Artificial Analysis leaderboard, surpassing even Fable 5. Its pricing is consistent with Opus 4.8, and it retains the option of a "fast mode" available at double the base model's cost. Demonstrating its advanced capabilities, Opus 5 successfully reconstructed a 3D FreeCAD model from a drawing of a machine part, even when it was intentionally not provided with direct viewing access, by developing its own computer vision pipeline. Anthropic has deliberately omitted training Opus 5 on cyber exploitation tasks, although its general capability improvements have led to a significant enhancement in identifying cybersecurity vulnerabilities, bringing it close to Mythos 5 in this area. However, it still lags behind Mythos 5 in the actual exploitation of these vulnerabilities. Anthropic has also released a prompting guide for Claude Opus 5 to assist users in leveraging its features effectively.
The release of Claude Opus 5 signifies a competitive push in the advanced AI model market, emphasizing enhanced capabilities alongside a more accessible price point. This strategy suggests a market dynamic where performance gains are being democratized, potentially accelerating AI adoption across various sectors. The model's proactive problem-solving, as demonstrated by its ability to infer and reconstruct complex data without explicit instructions, highlights a shift towards more autonomous AI agents. Anthropic's deliberate decision to refrain from training on cyber exploitation, while still improving vulnerability detection, reflects a responsible development approach that acknowledges the dual-use nature of powerful AI. This stance may mitigate regulatory risks and foster trust, positioning the company to navigate the evolving landscape of AI governance and ethical deployment over the next decade.
AI-generated to prompt reflection — not editorial opinion, not advice, not a statement of fact. How this works.