Skip to content

Benchmark

SpreadsheetBench 2

Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages. Higher is better.

Models measured
11
Scores recorded
12
Sources
3
IQ weight
0.06

Score distribution

24.7%
Mean
26.8%
Median
10.1%
Std dev
11.2%
10th percentile
34.9%
90th percentile
7.2%
Min
7.2%36.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.06 for chat models, measured over 11 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1ERNIE 5.5 Flash Preview· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim36.4%
2LongCat 2.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim34.9%
3Claude Opus 4.6scaffold: swe-agent, date: 2026-03-26· SpreadsheetBench 2 leaderboard data34.9%
4Kimi K3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim34.8%
5Kimi K3max effort · temp 1.0 · top_p 1.0· Reported by Moonshot AI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.34.8%
6GPT 5.2scaffold: swe-agent, date: 2026-03-26· SpreadsheetBench 2 leaderboard data26.8%
7Gemini 3.1 Pro Previewscaffold: swe-agent, date: 2026-03-26· SpreadsheetBench 2 leaderboard data23.7%
8Ling 3.0 Flash Fin· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim21.8%
9DeepSeek V3.2scaffold: swe-agent, date: 2026-03-26· SpreadsheetBench 2 leaderboard data15.6%
10Kimi K2.5scaffold: swe-agent, date: 2026-03-26· SpreadsheetBench 2 leaderboard data14.6%
11Qwen3.5 397B A17Bscaffold: swe-agent, date: 2026-03-26· SpreadsheetBench 2 leaderboard data11.2%
12MiniMax M2.5scaffold: swe-agent, date: 2026-03-26· SpreadsheetBench 2 leaderboard data7.2%

Data credits: SpreadsheetBench (Renmin University of China and contributors), spreadsheetbench.github.io. · Data by LLM Stats · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.