Wikibench

AIME 2026

math

Multilingual: no

All 30 problems from the 2026 American Invitational Mathematics Examination (AIME I and AIME II), testing olympiad-level mathematical reasoning with integer answers from 000-999. Used as an AI benchmark to evaluate large language models' ability to solve complex mathematical problems requiring multi-step logical deductions and structured symbolic reasoning.

Methodology

Imported from llm-stats public benchmark metadata. Modality: text. Max score: 1. Categories: math, reasoning. Language: en. Verified by llm-stats: no.

Leaderboard

Showing 12 of 12 results

  1. Kimi K2.6 self-reported llm-stats
    96.4%
  2. GLM-5.1 self-reported llm-stats
    95.3%
  3. Qwen3.6 Plus self-reported llm-stats
    95.3%
  4. MAI-Thinking-1 self-reported llm-stats
    94.5%
  5. Seed 2.0 Pro self-reported llm-stats
    94.2%
  6. Qwen3.6-27B self-reported llm-stats
    94.1%
  7. MAI-Code-1-Flash self-reported llm-stats
    92.5%
  8. Qwen3.5-397B-A17B self-reported llm-stats
    91.3%
  9. Gemma 4 31B self-reported llm-stats
    89.2%
  10. Gemma 4 26B-A4B self-reported llm-stats
    88.3%
  11. Gemma 4 E4B self-reported llm-stats
    42.5%
  12. Gemma 4 E2B self-reported llm-stats
    37.5%