GPT-4.1 answered 96% of 25 checkable tasks correctly; GPT-4o mini 92% at 13 times less.
Seonara model bench, 2026-10-29: 25 tasks x 1 run(s) at temperature 0, task set 2026-10-29.1, every answer scored by a program.
| Model | Correct | Change | dates-units | extraction | instructions | knowledge | reasoning | Median seconds | Cost per run |
|---|---|---|---|---|---|---|---|---|---|
| GPT-4.1 | 96% | new | 75% | 100% | 100% | 100% | 100% | 0.83 | $0.0028 |
| GPT-4o mini | 92% | new | 75% | 100% | 80% | 100% | 100% | 0.78 | $0.0002 |
| GPT-4.1 mini | 88% | new | 75% | 100% | 60% | 100% | 100% | 0.76 | $0.0006 |
What every model got wrong
- date-days (dates-units): How many days are there from 1 March 2024 to 1 March 2025? Answer with the number only.
Expected
365; gpt-4.1 answered366; gpt-4.1-mini answered366; gpt-4o-mini answered366.
What this cannot tell you
25 questions is a small sample: a one-question difference is 4 points, so models within a few points of each other are tied. Seconds are wall time from one PC, network included. Costs use the listed prices per million tokens on the day. The questions, the scoring code and every answer are published, so the run can be repeated.
Raw results: results.csv and methodology.md of the 2026-10-29 run, published with this scoreboard.