Claude Sonnet 4.5: The Best Coding Model? Beats GPT-5 & Gemini 2.5 Pro (Hands-On)

2025-10-11 14:288 min read

In this video, the presenter discusses the newly released Claudet 4.5 AI model, emphasizing its improved capabilities over previous versions, particularly in coding simulations. They showcase a 3D Rubik's cube simulation created using this version and elaborate on a cost of $9.34 for executing a task using kilo code. The presenter expresses confusion over the cost and discusses potential alternatives for using the model, such as signing up for kilo code for free credits or opting for paid access. They detail how to engage with the model, highlight its advantages over competitors, and mention their experience using similar coding extensions. In conclusion, they encourage viewers to subscribe for more content regarding the AI model and its applications.

Key Information

  • Claudet 4.5 Sonet is a new model that is reported to be slightly better than Claw 4.1 Sonet and Claw 4 Sonet.
  • A demonstration was conducted using a 3D Rubik's cube simulation to judge coding capabilities of the AI model.
  • The test resulted in a cost of $9.34 for a simple task, raising questions about whether this amount is accurate or a mistake.
  • Claudet 4.5 costs $3 per million input tokens and $15 per million output tokens, similar to previous models.
  • The new model boasts a 1 million context window, which is considered impressive.
  • A direct way to use the model is via the website claude.ai or through kilo code, which requires a paid subscription.
  • OpenRouter also offers flexibility in accessing the model, which can work with different Client and Rule codes.
  • The model has been highlighted to outperform other models like Claude Opus 4.1 and Claude Sonet 4 in various benchmarks.
  • Future videos will focus on creating more complex projects using Claude 4.5 and discussing additional features.

Timeline Analysis

Content Keywords

Claudet 4.5

Claudet 4.5 sonet is the latest AI model, slightly better than previous versions like Claw 4.1 and Claude 4. The creator shares a 3D Rubik's cube simulation test to evaluate its coding capabilities.

Kilo Code

The testing was conducted using Kilo Code, which cost $9.34 for a simple task. The system appeared to have some issues, which were later resolved.

Cost Structure

Claudet 4.5 operates on a cost-per-million input tokens basis ($3) and a charge of $15 per million output tokens. The model provides a million context window.

Comparison

Claudet 4.5 outperformed multiple AI models like Claude Opus 4.1 and GPD 5 across various domains including finance, law, and medicine.

Usage Options

Access Claudet 4.5 through direct web use at claude.ai or via the Kilo Code platform. Users can also explore different API providers for pricing benefits.

Performance Testing

Different benchmarks show Claudet 4.5 slightly exceeds other models, indicating superior performance in tasks requiring reasoning and coding.

User Experience

The creator shares positive experiences using the model, affirms its capabilities in coding tasks, and promotes upcoming content related to Claudet 4.5.

More video recommendations

Share to: