Android Bench 2.0

AI-assisted software engineering has seen the emergence of several benchmarks to measure the capabilities of LLMs. Android developers face specific challenges that aren't covered by existing benchmarks, so we created one that focuses on a north star of high quality Android development.

Long-horizon task results

Pass rate

28.0%

Avg completion rate

82.2%

Avg latency

7.9 h

Avg cost

$375.7

Per task results

Pass rate

22.7%

Avg completion rate

82.4%

Avg latency

22.2 h

Avg cost

$492.6

Per task results

Pass rate

19.3%

Avg completion rate

74.3%

Avg latency

8.6 h

Avg cost

$235.8

Per task results

Pass rate

16.7%

Avg completion rate

77.8%

Avg latency

27.0 h

Avg cost

$861.4

Per task results

Pass rate

14.0%

Avg completion rate

74.3%

Avg latency

47.2 h

Avg cost

$260.2

Per task results

Pass rate

12.0%

Avg completion rate

72.8%

Avg latency

66.2 h

Avg cost

$418.3

Per task results

Pass rate

8.0%

Avg completion rate

47.4%

Avg latency

12.1 h

Avg cost

$34.5

Per task results

Pass rate

7.3%

Avg completion rate

50.1%

Avg latency

9.9 h

Avg cost

$26.0

Per task results

Pass rate

6.7%

Avg completion rate

59.8%

Avg latency

14.7 h

Avg cost

$283.7

Per task results

Pass rate

4.7%

Avg completion rate

53.4%

Avg latency

5.1 h

Avg cost

$53.2

Per task results

1.0 task results

Track the latest AI model benchmarks, newly introduced agent architectures, and continuous performance evaluations on the platform. Stay updated with our routine methodology updates and release logs.

-

New updates • Sep 17thLaunched long-horizon tasks

-

New updates • Sep 17thNow benchmarking models using specialized agents

-

New agent results • Sep 17th

antigravity-sdk results now availableGemini 3.6 Flash, Gemini 3.7 Flash

-

New agent results • Sep 17th

pi results now availableGemini 3.6 Flash

-

New agent results • Sep 17th

codex results now availableGPT 5.6 Luna, GPT 5.6 Sol

-

New models • Aug 11thClaude Opus 5

-

New models • Aug 11thGPT 5.6 Sol, GPT 5.6 Luna, GPT 5.6 Terra

-

New models • Aug 11thKimi K3

-

New models • Aug 11thQwen3.8 Max

-

New models • Aug 11thGemini 3.6 Flash, Gemini 3.5 Flash Lite

-

Archived models • Aug 11thGemma 4 26B A4B IT