ArXivMath leaderboard
As of 16 Nov 2026, the top score on ArXivMath is 94.3% by Gemini 4 Carbon (Google), ahead of GPT-6.1 Sol at 92.8%, across 8 model configurations with a published result (MathArena). Research-level final-answer problems taken from arXiv papers of the last few months, refreshed monthly; the Overall table combines the current monthly sets. MathArena. Shown for reference; few models so far. ArXivMath is shown for reference only and does not count toward the BenchLeader Index.
Key facts (as of 16 Nov 2026)
- Gemini 4 Carbon holds the top ArXivMath score at 94.3% (MathArena, 16 Nov 2026).
- GPT-6.1 Sol is second on ArXivMath at 92.8% (MathArena, 16 Nov 2026).
- ArXivMath: Research-level final-answer problems taken from arXiv papers of the last few months, refreshed monthly; the Overall table combines the current monthly sets. MathArena. Shown for reference; few models so far (MathArena).
- 8 model configurations have a published ArXivMath result on BenchLeader (BenchLeader, 16 Nov 2026).
- ArXivMath does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).
Research-level final-answer problems taken from arXiv papers of the last few months, refreshed monthly; the Overall table combines the current monthly sets. MathArena. Shown for reference; few models so far.
- Published by
- MathArena
- Category
- Maths
- Index weight
- Reference only
- Models
- 8
- Data as of
- 16 Nov 2026
MIT — eth-sri/matharena.
- 1Gemini 4 Carbon (high)94.3%
- 2GPT-6.1 Sol (max)92.8%
- 3Gemini 4 Argon (high)84.4%
- 4Claude Opus 5.5 (high)82.6%
- 5Kimi K3.1 Pro (max)58.3%
- 6DeepSeek V4.1 Flash 1017 (max)42.5%
- 7Inkling 2 Small31.9%
- 8Muse Spark Open (xhigh)27.0%
| # | |||||
|---|---|---|---|---|---|
| 1 | 94.3% | 68.9 | $5.08 | 203k | |
| 2 | 92.8% | 66.6 | $0.960 | 64k | |
| 3 | 84.4% | 65.3 | $3.74 | 182k | |
| 4 | 82.6% | 66.0 | $1.93 | 52k | |
| 5 | 58.3% | 59.9 | $2.94 | 103k | |
| 6 | 42.5% | 57.7 | $0.330 | 179k | |
| 7 | 31.9% | 57.9 | $0.600 | 216k | |
| 8 | 27.0% | 53.5 | $1.03 | 86k |
AI reasoning and math benchmark leaderboard · All LLM benchmarks
Related benchmarks
Top head-to-heads on ArXivMath
Cite as: BenchLeader, “ArXivMath leaderboard”, data as of 16 Nov 2026, https://www.benchleader.com/benchmarks/matharena_arxivmath.
ArXivMath: questions
- What does ArXivMath measure?
- Research-level final-answer problems taken from arXiv papers of the last few months, refreshed monthly; the Overall table combines the current monthly sets. MathArena. Shown for reference; few models so far. Scores are reported in percent of tasks solved; higher is better.
- What is the top score on ArXivMath?
- As of 16 Nov 2026, the highest score on ArXivMath is 94.3% by Gemini 4 Carbon, ahead of GPT-6.1 Sol at 92.8%. Scores come from MathArena and are re-read daily.
- Which AI model leads ArXivMath?
- Gemini 4 Carbon leads ArXivMath with 94.3% as of 16 Nov 2026, ahead of GPT-6.1 Sol at 92.8%.
- How many models have ArXivMath results?
- 8 model configurations have a ArXivMath result on BenchLeader, all taken from MathArena.
- Who runs ArXivMath and how often is it updated?
- ArXivMath is published by MathArena. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does ArXivMath count toward the BenchLeader Index?
- No. ArXivMath is shown for reference but left out of the composite index.