DeepSeek-V4-Flash-Max: benchmarks, price and rank

Its headline results

The benchmarks that count most in the rating, with its place among the models measured the same way, and where each number came from.

• BrowseComp73.2%

  Questions whose answers are hidden deep on the web and take many searches and pages to track down.

  30th of 56. The maker's own number, from DeepSeek, via LLM Stats.

• MCP Atlas69%

  Multi-step tasks where the model has to pick and combine many real tools, connected through MCP, the standard way agents plug into apps and data.

  20th of 33. The maker's own number, from DeepSeek, via LLM Stats.

• SWE-Bench Pro52.6%

  Long changes to real codebases in several languages, often across many files, including code from private and copyleft projects that models are unlikely to have seen.

  41st of 50. The maker's own number, from DeepSeek, via LLM Stats.

• Humanity's Last Exam45.1%

  About 2,500 questions written by experts in more than a hundred fields to be as hard as they could make them, many needing a diagram or image.

  23rd of 74. The maker's own number, from DeepSeek, via LLM Stats.

• Toolathlon47.8%

  Long tasks across many real apps and their tools, such as email, calendars, spreadsheets, code repositories and online stores.

  23rd of 37. The maker's own number, from DeepSeek, via LLM Stats.

• IMO-AnswerBench88.4%

  Problems from the International Mathematical Olympiad, the hardest contest for school students, where the model has to reach the right final answer.

  5th of 18. The maker's own number, from DeepSeek, via LLM Stats.

• SWE-bench Verified79%

  Real bug reports and feature requests from popular open-source Python projects. The model has to change the code so the project's hidden tests pass.

  7th of 56. The maker's own number, from DeepSeek, via LLM Stats.

• GPQA Diamond88.1%

  Graduate-level questions in biology, physics and chemistry, written by PhDs so that searching the web does not give the answer away.

  The maker's own number, from DeepSeek, via LLM Stats.

• MMLU-Pro86.2%

  Twelve thousand multiple-choice questions across fourteen subjects, from law and medicine to engineering, with ten options each.

  11th of 56. The maker's own number, from DeepSeek, via LLM Stats.