Skip to content

Benchmark

KiloBench Terminal-Bench

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

Models measured
29
Scores recorded
29
Sources
1
IQ weight
1.76

Score distribution

60.9%
Mean
60.7%
Median
13.6%
Std dev
47.6%
10th percentile
75.3%
90th percentile
19.1%
Min
19.1%79.3%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.76 for chat models, measured over 29 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT-6 Astraterminal-bench· Kilo AI Gateway data79.3%
2Claude Fable 5.1terminal-bench· Kilo AI Gateway data76.2%
3GPT-5.6 Solterminal-bench· Kilo AI Gateway data76.2%
4DeepSeek V4.1 Flashterminal-bench· Kilo AI Gateway data75.3%
5Gemini 3.8 Flashterminal-bench· Kilo AI Gateway data75.3%
6GPT 5.5terminal-bench· Kilo AI Gateway data74.2%
7Grok 4.6terminal-bench· Kilo AI Gateway data73%
8Kimi K3terminal-bench· Kilo AI Gateway data72.8%
9Claude Opus 5terminal-bench· Kilo AI Gateway data71.5%
10Claude Fable 5terminal-bench· Kilo AI Gateway data71%
11Grok 4.5terminal-bench· Kilo AI Gateway data70.8%
12Claude Opus 4.7terminal-bench· Kilo AI Gateway data70.1%
13Claude Opus 4.8terminal-bench· Kilo AI Gateway data67.6%
14Gemini 3.5 Flashterminal-bench· Kilo AI Gateway data64.7%
15Kimi K2.7 Codeterminal-bench· Kilo AI Gateway data60.7%
16Muse Spark 1.1terminal-bench· Kilo AI Gateway data59.8%
17Claude Sonnet 5terminal-bench· Kilo AI Gateway data59.5%
18Claude Sonnet 4.6terminal-bench· Kilo AI Gateway data55.1%
19Qwen3.7 Maxterminal-bench· Kilo AI Gateway data54.6%
20Kimi K2.6terminal-bench· Kilo AI Gateway data54.4%
21GLM 5.2terminal-bench· Kilo AI Gateway data53%
22Muse Spark 1.2terminal-bench· Kilo AI Gateway data51.5%
23GLM 5.1terminal-bench· Kilo AI Gateway data49.4%
24Gemini 3.6 Flashterminal-bench· Kilo AI Gateway data47.9%
25MiMo V2.5 Proterminal-bench· Kilo AI Gateway data47.6%
26MiniMax M3terminal-bench· Kilo AI Gateway data47.6%
27DeepSeek V4 Proterminal-bench· Kilo AI Gateway data44%
28Inkling FP4terminal-bench· Kilo AI Gateway data43.6%
29Nemotron 3 Ultraterminal-bench· Kilo AI Gateway data19.1%