Skip to content

Benchmark

Meta Internal Coding Bench

Coding benchmarks compare models on programming tasks. Scores are percentages. Higher is better.

Models measured
5
Scores recorded
6
Sources
2
IQ weight
0.98

Score distribution

69.7%
Mean
70.6%
Median
5%
Std dev
65.4%
10th percentile
79.4%
90th percentile
63.9%
Min
63.9%79.4%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.98 for chat models, measured over 5 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Claude Opus 5max· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.79.4%
2Muse Spark 1.2· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim70.6%
3Muse Spark 1.2Meta 官方发布图· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.70.6%
4Muse Spark 1.1Meta 官方发布图· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.68.3%
5GPT-5.6 Terramax· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.65.4%
6Gemini 3.6 Flashhigh· Reported by Meta· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.63.9%

Data credits: 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. · Data by LLM Stats