# Scale Labs SEAL Math (SEAL Math)

> Scale’s SEAL math board on GSM1k, a new set of grade-school word problems written to mirror GSM8K without its contamination. BenchLM shows it as display-only reference data, excluded from rankings. Scale marked this board deprecated as of January 2025 and no longer adds models.

Canonical page: https://benchlm.ai/benchmarks/scale-math

- Category: [Mathematics](/math)
- Last updated: November 12, 2026 snapshot

## About SEAL Math

- Year: 2026
- Tasks: About 1,000 problems (GSM1k)
- Format: Accuracy (%)
- Difficulty: External agent and model evaluation
- Paper: [Scale Labs leaderboard](https://labs.scale.com/leaderboard/math)

GSM1k contains about 1,000 problems at roughly a fifth-grade level. The score is accuracy in percent. BenchLM mirrors 16 published rows captured on November 12, 2026 snapshot.

SEAL Math is currently displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.

## Leaderboard (16 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [Claude 3.5 Sonnet (June 2024)](https://labs.scale.com/leaderboard/math) | Anthropic | 96.60% |
| 2 | [GPT-4o (August 2024)](https://labs.scale.com/leaderboard/math) | OpenAI | 95.68% |
| 3 | [Llama 3.1 405B Instruct](https://labs.scale.com/leaderboard/math) | Meta | 95.60% |
| 4 | [Claude 3 Opus](/models/claude-3-opus) | Anthropic | 95.19% |
| 5 | [GPT-4 Turbo Preview](https://labs.scale.com/leaderboard/math) | OpenAI | 95.10% |
| 6 | [GPT-4o (May 2024)](https://labs.scale.com/leaderboard/math) | OpenAI | 94.85% |
| 7 | [Gemini 1.5 Pro (August 2024)](https://labs.scale.com/leaderboard/math) | Google | 94.69% |
| 8 | [Mistral Large 2](/models/mistral-large-2) | Mistral | 93.94% |
| 9 | [Claude 3 Sonnet](https://labs.scale.com/leaderboard/math) | Anthropic | 93.28% |
| 10 | [Gemini 1.5 Pro (May 2024)](https://labs.scale.com/leaderboard/math) | Google | 92.28% |
| 11 | [Gemini 1.5 Pro (April 2024)](https://labs.scale.com/leaderboard/math) | Google | 90.54% |
| 12 | [Gemini 1.5 Flash](/models/gemini-1-5-flash) | Google | 90.12% |
| 13 | [Llama 3 70B Instruct](https://labs.scale.com/leaderboard/math) | Meta | 90.12% |
| 14 | [Mistral Large](https://labs.scale.com/leaderboard/math) | Mistral | 87.47% |
| 15 | [Gemini 1.0 Pro](/models/gemini-1-0-pro) | Google | 79.83% |
| 16 | [CodeLlama 34B Instruct](https://labs.scale.com/leaderboard/math) | Meta | 37.51% |

## FAQ

### What does SEAL Math measure?

Scale’s SEAL math board on GSM1k, a new set of grade-school word problems written to mirror GSM8K without its contamination. BenchLM shows it as display-only reference data, excluded from rankings. Scale marked this board deprecated as of January 2025 and no longer adds models.

### Which model leads the published SEAL Math snapshot?

Claude 3.5 Sonnet (June 2024) currently leads the published SEAL Math snapshot with a score of 96.60%.

### How many models are evaluated on SEAL Math?

The November 12, 2026 snapshot contains 16 AI models.

### Does SEAL Math affect BenchLM's overall score?

Not directly. SEAL Math is still displayed on BenchLM for reference, but it is excluded from the weighted scoring formula.