PHREEQC-MCQ-200 (Raw Output): leaderboard
PHREEQC-MCQ-200 (2026) agents answering simulator-grounded aqueous-geochemistry questions when the PHREEQC output is returned inline, truncated to its first and last 100,000 characters, instead of through a table of contents. Accuracy in percent over 200 four-option items, higher is better.
Metric: Accuracy (%; 200 four-option questions from 21 validated PHREEQC aqueous-geochemistry simulation scenarios; the custom agent writes a PHREEQC input, runs the simulator and reads its output within 24 steps, then writes one answer letter; a missing or unparseable answer counts as wrong; simulator output inlined and head/tail truncated to 100,000 characters, Agent-Raw100k). Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.6 | 83.5 | #161 |
| 2 | Claude Opus 4.6 | 83 | #102 |
| 3 | GPT-5.4 | 70 | #123 |
| 4 | GPT-5.1 | 63.5 | #229 |
| 5 | GPT-5.2 | 61.5 | #176 |
| 6 | Gemini 3 Flash (Preview) | 47 | #144 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=phreeqc-mcq-200-raw-output · How It Works · Data refreshed daily, snapshot 2026-11-13.