Benchmark
Terminal-Bench-Science 0.1 Evaluation Results
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages.
All agentic benchmarksScores from Harbor Hub Leaderboards.
- Models measured
- 14
- Scores recorded
- 14
- Sources
- 1
- IQ weight
- 2.42
Score distribution
- 17.9%
- Mean
- 10.5%
- Median
- 20.1%
- Std dev
- 3.8%
- 10th percentile
- 63.3%
- 90th percentile
- 3.3%
- Min
This benchmark has a weight of 2.42 for chat models, measured over 14 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | |
|---|---|---|
| 1 | GPT-6 Astrav0-1-eval· Harbor Hub Leaderboards | 68.1% |
| 2 | Claude Opus 5.5v0-1-eval· Harbor Hub Leaderboards | 63.3% |
| 3 | GPT-5.6 Solv0-1-eval· Harbor Hub Leaderboards | 22.4% |
| 4 | DeepSeek V4.1 Flashv0-1-eval· Harbor Hub Leaderboards | 15.7% |
| 5 | Grok 4.7v0-1-eval· Harbor Hub Leaderboards | 14.3% |
| 6 | Gemini 3.8 Flashv0-1-eval· Harbor Hub Leaderboards | 12.4% |
| 7 | Claude Opus 4.8v0-1-eval· Harbor Hub Leaderboards | 10.5% |
| 8 | GPT-5.6 Terrav0-1-eval· Harbor Hub Leaderboards | 8.6% |
| 9 | GLM 5.3v0-1-eval· Harbor Hub Leaderboards | 8.1% |
| 10 | Grok 4.6v0-1-eval· Harbor Hub Leaderboards | 7.1% |
| 11 | Kimi K3v0-1-eval· Harbor Hub Leaderboards | 7.1% |
| 12 | Gemini 3.7 Flashv0-1-eval· Harbor Hub Leaderboards | 5.7% |
| 13 | DeepSeek V4 Prov0-1-eval· Harbor Hub Leaderboards | 3.8% |
| 14 | GPT-5.6 Lunav0-1-eval· Harbor Hub Leaderboards | 3.3% |
Tokens per task, input and output together, as measured by the source. Comparable within this benchmark only.