Skip to content

Next evaluationfor solversof agents

NextEval

NextEval is an open research project on derivative-free optimization with language-model agents under a fixed budget of function evaluations: a benchmark that scores coding agents against COBYQA, and a solver in which an agent chooses the numerical methods.

/benchResults

NextEval Bench

A benchmark that scores coding agents on derivative-free optimization problems against the COBYQA solver.

Open the leaderboard
/solverRelease: coming soon

NextEval Solver 1.0

A derivative-free optimizer in which a language-model agent chooses numerical-method campaigns, with a textbook autopilot as the fixed-policy fallback.

Read about the solver

Bench at a glance

Full leaderboard

Last updated: 2026-11-06 · snapshot web-ffdd187d

First six of 45 agents by pair score against COBYQA (4-task mean). COBYQA = 0.500; above means ahead of COBYQA. The beat-COBYQA rate is the share of cases in which the agent reaches every tolerance no later than COBYQA and at least one strictly earlier.

First six agents of the NextEval Bench leaderboard
RankAgentHarnessFrom COBYQA (0.500)Pair scoreBeat rate
1GPT-6.1 Sol (high)Codex0.72638.0%
2GPT-6 Astra (high)Codex0.70637.2%
3Opus 5.5 (high)Claude Code0.69137.8%
4GPT-6.1 Sol (medium)Codex0.69033.8%
5Fable 5.1 (high)Claude Code0.68034.5%
6Hy4 (high)Claude Code0.67434.1%

Views: Leaderboard · Matrix · Waffle · Profiles · Pareto · Methodology