Scale Labs SEAL Korean (SEAL Korean)
We show this table for reference; we do not rank on it.
Scale's SEAL leaderboard for Korean: expert raters judge how well models answer about 1,000 prompts written by native Korean speakers. BenchLM shows it as display-only reference data, excluded from rankings. Scale marked this board deprecated as of March 2025 and no longer adds models.
Scale score on SEAL Korean — November 12, 2026 snapshot
We mirror the published scale score view for SEAL Korean. o1-preview leads the public snapshot at 66.43, followed by Claude 3.7 Sonnet (February 2025) (64.93) and GPT-4o (May 2024) (64.58). We do not use these results to rank models overall.
o1-preview
OpenAI
Claude 3.7 Sonnet (February 2025)
Anthropic
GPT-4o (May 2024)
OpenAI
19 modelsCross-domainCurrentDisplay onlyUpdated November 12, 2026 snapshot
Scale score table (19 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | o1-previewOpenAI | 66.43 | Not reported | Closed |
| 2 | Claude 3.7 Sonnet (February 2025)Anthropic | 64.93 | Not reported | Unknown |
| 3 | GPT-4o (May 2024)OpenAI | 64.58 | Not reported | Unknown |
| 4 | GPT-4.5 Preview (February 2025)OpenAI | 63.76 | Not reported | Unknown |
| 5 | GPT-4 Turbo PreviewOpenAI | 60.76 | Not reported | Unknown |
| 6 | Gemini 1.5 Pro (August 27, 2024)Google | 60.28 | Not reported | Unknown |
| 7 | GPT-4o (August 2024)OpenAI | 59.93 | Not reported | Unknown |
| 8 | Claude 3.5 Sonnet (June 2024)Anthropic | 59.38 | Not reported | Unknown |
| 9 | Gemini 2.0 FlashGoogle | 55.60 | Not reported | Closed |
| 10 | Claude 3 SonnetAnthropic | 54.17 | Not reported | Unknown |
| 11 | Claude 3 OpusAnthropic | 52.78 | Not reported | Closed |
| 12 | GPT-4o miniOpenAI | 51.74 | Not reported | Closed |
| 13 | GPT-4OpenAI | 51.39 | Not reported | Unknown |
| 14 | Mistral Large 2Mistral | 50.35 | Not reported | Closed |
| 15 | Llama 3.1 405B InstructMeta | 50.35 | Not reported | Unknown |
| 16 | Gemini 1.5 Pro (May 2024)Google | 40.42 | Not reported | Unknown |
| 17 | Llama 3.1 70B InstructMeta | 37.23 | Not reported | Unknown |
| 18 | Command R+Cohere | 30.21 | Not reported | Unknown |
| 19 | Llama 3.1 8B InstructMeta | 17.42 | Not reported | Unknown |
How to read this leaderboard
Compare the published configurations as complete evaluation systems. The source can combine a base model, agent scaffold, tools, budget, and inference setting in each result.
Operator receipt: 19 sourced rows are currently displayable on this page; the leading published row is o1-preview at 66.43.
Honest limit: This Scale table is display-only context, not benchmark provenance or a weighted model-only comparison. Scale marked this board deprecated as of March 2025 and no longer adds models.
How BenchLM shows SEAL Korean
BenchLM mirrors 19 published rows from Scale Labs’ public SEAL Korean leaderboard, captured on November 12, 2026 snapshot.
The table is display only. It is useful context for a published agent or model configuration, but it does not enter BenchLM’s overall or category rankings. Scale marked this board deprecated as of March 2025 and no longer adds models.
Snapshot
The published SEAL Korean snapshot places o1-preview first at 66.43. The third row is 1.85 score units behind. The broader top-10 range is 12.26 score units, so the table still separates the published systems.
19 models have been evaluated on SEAL Korean. The benchmark falls in the Cross-domain category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. SEAL Korean is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About SEAL Korean
Year
2026
Tasks
About 1,000 prompts per language
Format
Expert-rated score
Difficulty
External agent and model evaluation
Single-turn prompts across general and culturally specific scenarios, rated by native-speaking experts. Scale publishes this board as a 0 to 100 score rather than a rating. BenchLM mirrors 19 published rows captured on November 12, 2026 snapshot.
Freshness and provenance
Version
SEAL Korean 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does SEAL Korean measure?
Scale's SEAL leaderboard for Korean: expert raters judge how well models answer about 1,000 prompts written by native Korean speakers. BenchLM shows it as display-only reference data, excluded from rankings. Scale marked this board deprecated as of March 2025 and no longer adds models.
Which model leads the published SEAL Korean snapshot?
o1-preview currently leads the published SEAL Korean snapshot with 66.43 scale score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on SEAL Korean?
The November 12, 2026 snapshot snapshot contains 19 AI models.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.