Interaction BenchInteraction Bench
v0.1 preview
LeaderboardEnvironmentsTasksPipelineMethodologyDocsBlogAboutContributors
Github
OpenEMR

OpenEMR

Business & finance
Electronic health records and practice management
Githubopenemr/openemr5.5k
Tech stack
PHPJavaScriptMySQL
Languages
PHP88.6%
JavaScript2.8%
XSLT2.3%
HTML1.8%
Shell1.5%
Twig1.5%
Other1.5%
Available tools
Browser
Terminal
Network
Resume an open calendar after being idle
Resume an open calendar after being idle

Tasks

All 1 task
Resume an open calendar after being idle
#14233 · Base 87e84e9
After lunch, continue checking appointments in the calendar I left open.
33%

Model results

Full leaderboard
Model
Harness
Claude
Claude Opus 5.5
Claude Code
2.6M in · 18.1k out · 2.7M total
Claude Code · 42 steps · $1.53 · 7m 00s · 2.7M tokens
Claude Code
Claude Code
42
$1.53
7m 00s
2.6M
18.1k
2.7M
OpenAI
GPT-6 Astra
Codex
707.2k in · 6.2k out · 713.4k total
Codex · 20 steps · $1.66 · 10m 08s · 713.4k tokens
Codex
Codex
20
$1.66
10m 08s
707.2k
6.2k
713.4k
DeepSeek
DeepSeek Flash
DSH
37.2M in · 85.8k out · 37.3M total
DSH · 204 steps · $0.21 · 26m 41s · 37.3M tokens
DSH
204
$0.21
26m 41s
37.2M
85.8k
37.3M
Z.ai
GLM-5.3 Flash
ZCode
35.7M in · 78.7k out · 35.7M total
ZCode · 271 steps · $1.14 · 1h 05m · 35.7M tokens
ZCode
271
$1.14
1h 05m
35.7M
78.7k
35.7M
Qwen
Qwen3.8 Max
Codex
24.0M in · 83.8k out · 24.1M total
Codex · 215 steps · $7.05 · 1h 02m · 24.1M tokens
Codex
Codex
215
$7.05
1h 02m
24.0M
83.8k
24.1M
MoonshotAI
Kimi K3
Kimi Code
17.4M in · 98.3k out · 17.5M total
Kimi Code · 0 steps · $7.91 · 1h 05m · 17.5M tokens
Kimi
Kimi Code
0
$7.91
1h 05m
17.4M
98.3k
17.5M
Interaction BenchInteraction Bench

Benchmarked on what users feel.

// results

LeaderboardEnvironmentsTasks

// how it works

PipelineMethodologyDocsBlog

// project

AboutContributorsContact usGitHubReport an issue
$ ibench --release 0.1 --statuspreview · built by EvalSys