AI tool · AI models and chat
Claude
AI toolAI models and chat
Anthropic's premium model for engineering and independent tasks with a long context.
What it is
Long context and working with documents.
Claude 4.5 Sonnet, released in September/October 2025, confirmed its reputation as a premium model for engineering and independent tasks. On the SWE-bench Verified test it reached 77.2% (82% with parallel computation) and so beat GPT-5 in programming. It can also do "Computer Use", meaning it controls the mouse and keyboard and moves around the computer's graphical environment; on the OSWorld test it reached 61.4%. It is built for "long-horizon tasks" and holds focus on a single goal for 30 or more hours of continuous operation.
What it does
What it does
- Top-tier programming: 77.2% on the SWE-bench Verified test (82% with parallel computation)
- Computer Use: 61.4% on the OSWorld test, controlling the mouse and keyboard in a graphical interface
- Long endurance: 30+ hours of continuous work on a single goal
- 200K token context as standard, up to 1M through special access
- Optimised for changing and migrating old code
Good for
- Complex programming, code changes and migrations between frameworks
- Analysing extensive texts and the whole code of projects
- Independent work in a graphical interface and designing system architectures
Not for
- Fast, less structured creative exploration of ideas (Gemini is a better fit)
Our verdict
Our choice for engineering and long-running tasks. In programming it beats GPT-5, and the combination of Computer Use with 30+ hours of endurance makes it the strongest agent for complex projects.
Why we use it
We use Claude 4.5 Sonnet for long context and working with documents. The model excels at programming, and at changing and migrating old code. It suits analysing extensive texts, system designs and working with complex inputs. Computer Use enables independent work in graphical interfaces, and the long endurance matters on complex projects.
Use cases
Use cases
Coding and code changes
Extensive changes to old code and migrations between frameworks, using its strength in programming and long endurance.
Document analysis
Analysing extensive texts and documents using a 200K to 1M token context, so it can handle whole collections of documents.
Independent work in applications
Independent work with graphical interfaces using Computer Use: moving around in applications and working with them.
System design
Designing and debugging complex system architectures, using long endurance and system capabilities.
AI models and chat
More tools in this category
Start here
15 minutes is enough for you to know whether it makes sense.
You tell us what is bothering you. We tell you whether we can help, roughly what it costs and where I would start in your place. No presentation.
- jakub@transformuj.ai
- Phone
- +420 773 070 090
- Address
- Legerova 1820/39, Praha 2