OpenAI's ChatGPT 5.4 Beats Claude Opus 4.6 and Gemini 3.1 Pro in a Habr Comparison

Habr has published a large practical comparison of three flagship models: Gemini 3.1 Pro, ChatGPT 5.4 and Claude Opus 4.6. The author tested not abstract benchmarks but ordinary everyday tasks, from writing a short story and compressing a PDF to math and a Python application, and by total score unexpectedly put ChatGPT in the lead.

How the models were compared

The test had four scenarios that users bring to neural networks every day. First, the models were asked to write a humorous fantasy story in three chapters. Then they were given a PDF with a practical assignment and asked to produce a condensed but usable summary without losing key information. Next came a block of four math problems, and the final test was building a desktop Python application: an engineering calculator with a GUI and a built-in Snake game.

The scoring was as practical as possible. The author graded the text and code tasks on a three-point scale, while the math stage gave up to four points, one for each correctly solved problem. For the first time, he also put the cost of each request in rubles into the table. As a result, the comparison covers not only the quality of the answer but also the price of getting it. The maximum under this scheme is 13 points, and the combination of score and spending became the main criterion for the final verdict.

Who won each stage

In the first stage, ChatGPT slipped slightly because of a chapter-numbering glitch and received 2.5 points, while Gemini and Claude each took the maximum 3 points. In the second round the picture reversed: ChatGPT compressed the PDF best and kept the important details, while Gemini and Claude, in the author's view, cut the text too aggressively and lost some necessary information. The math block was even for all three, but in programming differences showed up again, this time not in theory but in the working result.

• Text generation: Gemini 3.1 Pro, 3 points for 20 rubles; Claude Opus 4.6, 3 points for 68 rubles; ChatGPT 5.4, 2.5 points for 25 rubles.
• PDF compression: ChatGPT 5.4 got 3 points for 24 rubles; Gemini 3.1 Pro and Claude Opus 4.6 got 2 points each, for 16 and 38 rubles respectively.
• Math: all three models solved every problem, but ChatGPT 5.4 was cheaper, at 15 rubles versus 22 for Gemini and 29 for Claude.
• Programming: ChatGPT 5.4 got 3 points for a working calculator and Snake game; Gemini 3.1 Pro got 2.5 points because of faulty key capture in the game; Claude Opus 4.6 got 2 points because of an error when dividing decimal numbers.

"The result is clear: ChatGPT 5.4 won."

Price and trade-offs

The final table is revealing. ChatGPT 5.4 scored 11.5 points and spent 112 rubles. Gemini 3.1 Pro finished the test with 10.5 points and total costs of 87 rubles, making it the most economical option. Claude Opus 4.6 scored 10 points but cost 208 rubles, almost twice as much as ChatGPT and more than twice as much as Gemini. On price alone, Google leads; on the balance of quality and cost, OpenAI has the edge.

The comparison does not claim to be a universal academic benchmark. The author compares the models on a narrow set of everyday tasks and in places relies on his own editorial judgment, especially on text style or interface convenience. That is also what makes it useful: it shows not lab records but how the models behave in practical work. In this sample, Gemini looks like a rational budget option, Claude like an expensive and uneven one, and ChatGPT like the most consistent compromise.

What this means

If you want a single model for a broad range of everyday tasks, ChatGPT 5.4 comes out ahead in this comparison: it is not the best at everything, but it more often delivers the most consistent result for reasonable money. Gemini 3.1 Pro remains a strong alternative for those watching their budget closely, and after this test Claude Opus 4.6 looks like a less attractive choice than before.