Skip to content

Benchmark

SWE Atlas - Codebase QnA

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

Models measured
7
Scores recorded
7
Sources
1
IQ weight
1.85

Score distribution

50.1%
Mean
58%
Median
10.7%
Std dev
37.9%
10th percentile
59.4%
90th percentile
31.8%
Min
31.8%59.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.85 for chat models, measured over 7 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Mistral Large 4· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim59.4%
2Muse Spark 1.3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim59.4%
3LongCat 2.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim58.3%
4Muse Spark Open· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim58%
5Laguna S 2.1· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim46.2%
6MiniMax M3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim37.9%
7Beam 1.0 501B A23B· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim31.8%

Data credits: Data by LLM Stats