# Scale Labs SEAL Instruction Following (SEAL Instruction Following)

> Scale’s SEAL board for precise instruction following on 1,054 private prompts with detailed, specific commands. 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-instruction-following

- Category: [Instruction Following](/instruction-following)
- Last updated: November 12, 2026 snapshot

## About SEAL Instruction Following

- Year: 2026
- Tasks: 1,054 prompts
- Format: Expert-rated score
- Difficulty: External agent and model evaluation
- Paper: [Scale Labs leaderboard](https://labs.scale.com/leaderboard/instruction_following)

Unlike IFEval, the prompts are not limited to instructions a program can check; expert raters judge whether each reply follows every instruction. BenchLM mirrors 19 published rows captured on November 12, 2026 snapshot.

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

## Leaderboard (19 models)

| Rank | Model | Creator | Score |
|------|-------|---------|-------|
| 1 | [o1 (December 2024)](https://labs.scale.com/leaderboard/instruction_following) | OpenAI | 91.96% |
| 2 | [DeepSeek-R1](/models/deepseek-r1) | DeepSeek | 87.75% |
| 3 | [o1-preview](/models/o1-preview) | OpenAI | 86.58% |
| 4 | [Gemini 2.0 Flash Experimental (December 2024)](https://labs.scale.com/leaderboard/instruction_following) | Google | 86.58% |
| 5 | [Claude 3.5 Sonnet (June 2024)](https://labs.scale.com/leaderboard/instruction_following) | Anthropic | 85.96% |
| 6 | [GPT-4o (May 2024)](https://labs.scale.com/leaderboard/instruction_following) | OpenAI | 85.29% |
| 7 | [Llama 3.1 405B Instruct](https://labs.scale.com/leaderboard/instruction_following) | Meta | 84.85% |
| 8 | [Gemini 1.5 Pro (August 27, 2024)](https://labs.scale.com/leaderboard/instruction_following) | Google | 84.17% |
| 9 | [GPT-4 Turbo Preview](https://labs.scale.com/leaderboard/instruction_following) | OpenAI | 83.19% |
| 10 | [Mistral Large 2](/models/mistral-large-2) | Mistral | 82.81% |
| 11 | [GPT-4o (November 2024)](https://labs.scale.com/leaderboard/instruction_following) | OpenAI | 82.52% |
| 12 | [DeepSeek V3](/models/deepseek-v3) | DeepSeek | 82.34% |
| 13 | [Llama 3.2 90B Vision Instruct](https://labs.scale.com/leaderboard/instruction_following) | Meta | 82.07% |
| 14 | [Llama 3 70B Instruct](https://labs.scale.com/leaderboard/instruction_following) | Meta | 81.17% |
| 15 | [GPT-4o (August 2024)](https://labs.scale.com/leaderboard/instruction_following) | OpenAI | 80.17% |
| 16 | [Claude 3 Opus](/models/claude-3-opus) | Anthropic | 80.12% |
| 17 | [Mistral Large](https://labs.scale.com/leaderboard/instruction_following) | Mistral | 79.89% |
| 18 | [GPT-4 (November 2024)](https://labs.scale.com/leaderboard/instruction_following) | OpenAI | 79.50% |
| 19 | [Gemini 1.5 Pro (May 2024)](https://labs.scale.com/leaderboard/instruction_following) | Google | 79.37% |

## FAQ

### What does SEAL Instruction Following measure?

Scale’s SEAL board for precise instruction following on 1,054 private prompts with detailed, specific commands. 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 Instruction Following snapshot?

o1 (December 2024) currently leads the published SEAL Instruction Following snapshot with a score of 91.96%.

### How many models are evaluated on SEAL Instruction Following?

The November 12, 2026 snapshot contains 19 AI models.

### Does SEAL Instruction Following affect BenchLM's overall score?

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