TimeSage-MT - Code Correctness: leaderboard

Metric: Code correctness (%): half for code that executes cleanly, half for the share of reference output numbers recovered within tolerance, averaged over the applicable tasks; Code-Enabled Reasoning setting (the model writes Python, a sandbox runs it, the model answers from stdout; 240 multi-turn time-series analysis tasks over 8 domains), temperature 0, reasoning effort high where supported and medium for GPT-5.3-Codex; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 6 models tracked.

Top models

#ModelScore
1Claude Sonnet 4.678.51
2GLM-5.178.06
3GPT-5.3 Codex77.79
4Qwen 3.5 122B A10B72.89
5Gemini 3 Flash (Preview)72.39
6MiniMax-M2.771.74

Interactive version: theaggregate.ai/benchmark?slug=timesage-mt-code-correctness · How It Works · Data refreshed daily, snapshot 2026-09-29.