AI startup Anthropic has launched its most powerful AI model so far, Claude Opus 4.8. This new model has been introduced with better reasoning, coding and the ability to work without human help for a long time. The special thing is that Elon Musk, owner of rival company XAI, has also openly praised it. Musk described Anthropic’s work as “fantastic” on social media, after which it is being discussed a lot in the tech industry.
New model came into discussion due to praise of Elon Musk
After the launch of Cloud Opus 4.8, the most talked about thing is Elon Musk’s reaction. Musk’s company XAI’s Grok directly competes with AI models like Cloud, ChatGPT and Gemini. In such a situation, it is very rare to publicly praise the product of a rival company. Musk responded on social media by writing, “Great job” for Anthropic. After this, the question started being raised in the tech world that what is so special about this model, which attracted the attention of rival companies. This reaction of Musk is also being considered a sign of healthy competition in the AI industry.
Big upgrade in reasoning, coding and self-awareness
According to Anthropic, several significant improvements have been made in Cloud Opus 4.8. The company claims that this model can take more accurate decisions by understanding the situation and available information better than before. Apart from this, its self-awareness capability has also been improved, so that it can recognize its limitations and avoid making false claims. The model is also designed to handle long and complex projects without reducing performance. The company says that this AI tool will prove to be more capable than before in tasks like coding, analysis and problem solving, which will benefit both professionals and developers.
| Benchmark/Capacity | cloud opus 4.8 | cloud opus 4.7 | GPT-5.5 | Gemini 3.1 Pro |
| Agentic Coding (SWE-Bench Pro) | 69.20% | 64.30% | 58.60% | 54.20% |
| Agentic Terminal Coding (Terminal-Bench 2.1) | 74.60% | 66.10% | 78.20% | 70.30% |
| Multidisciplinary Reasoning (Humanity’s Last Exam – Without Tools) | 49.80% | 46.90% | 41.40% | 44.40% |
| Multidisciplinary Reasoning (Humanity’s Last Exam – with Tools) | 57.90% | 54.70% | 52.20% | 51.40% |
| Agentic Computer Use (OSWorld-Verified) | 83.40% | 82.80% | 78.70% | 76.20% |
| Knowledge Work (GDPval-AA) | 1890 | 1753 | 1769 | 1314 |
| Agentic Financial Analysis (Finance Agent v2) | 53.90% | 51.50% | 51.80% | 43.00% |
Great performance in benchmark test, low price too
A major feature of Cloud Opus 4.8 is its price. While companies usually offer newer and more capable AI models with expensive subscription plans, Anthropic has made it available at the same price as the older version. According to benchmark data shared by the company, the model achieved a score of 69.2 percent in the SWE-Bench Pro test. It got 57.9 percent marks in the Humanities Last Exam and 53.9 percent marks in the Financial Analysis Test. The company claims that this model accepts its uncertainty instead of guessing when the answer to the question is not known. This is why it is being seen as a more reliable AI model.
Also read: Bluetooth was named after which king? More than half the public does not know the correct answer
Leave a Reply