ORAgentBench - Medium Feasibility: leaderboard

Metric: Mean feasibility (0-100) on the 41 medium tasks: share of submissions that satisfy the schema, domains and hard operational constraints; end-to-end operations research tasks packaged with multi-file operational data, configuration and a submission schema; the agent writes and runs solver code in an isolated Harbor environment (45-minute interaction budget, 5-minute solve limit) and hidden validators score schema and hard-constraint feasibility and normalized objective quality; printed 0-1 and x100 here; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 14 models tracked.

Top models

#ModelScore
1GPT-5.4 (High)68
2GPT-5.3 Codex (High)68
3Claude Opus 4.659
4GPT-5.4 Mini (High)54
5Claude Sonnet 4.651
6Kimi K2.651
7DeepSeek V4 Pro46
8GLM-5.146
9MiMo-V2.5-Pro46
10DeepSeek V4 Flash41
11GLM-534
12Qwen 3.6 Plus29
13Qwen 3.5 Plus24
14MiniMax-M2.720

Interactive version: theaggregate.ai/benchmark?slug=oragentbench-medium-feasibility · How It Works · Data refreshed daily, snapshot 2026-09-29.