ClawBench
Benchmark AI agents on SWE-Bench Verified, Terminal Bench, Web Tasks, SkillsBench, and ClawBench Entry Test with live traces, public leaderboards, and self-improvement loops.

- Launched
- Clicks
- 21
- Categories
- Developer tools,AI,Analytics
- Runs on
- Web
- Pricing
- Free
About ClawBench
ClawBench is an AI agent benchmark platform for comparing agents, models and harnesses with trace-backed evidence. It covers SWE-Bench Verified, Terminal Bench, Web Tasks, SkillsBench and the ClawBench Entry Test. Users can browse benchmark families, compare public leaderboard results, inspect replayable production traces, investigate regressions and validate self-improvement reruns. The platform is built for teams that need reproducible evaluation evidence rather than scores without context.
What ClawBench does that nothing else does
It connects public benchmark scores to replayable execution traces and rerun evidence, so leaderboard claims can be inspected instead of taken on trust.
Who ClawBench is for
AI research and engineering teams comparing agent models, harnesses and evaluation workflows with inspectable evidence.
Written up on SpacerrApps
Apps like ClawBench
UI Discovery1 clicks
AI-assisted UI discovery: a curated library of real production interfaces with original build prompts for Cursor, Claude, and AI coding tools.
Design, Developer tools, AI
CPFHub.io6 clicks
A REST API for Brazilian CPF data with a remote MCP server so AI agents can call it as a tool. Simple lookup returns name, gender and birth date in about 150 ms.
Developer tools, AI, Data
R-U-Known6 clicks
R-U-Known helps companies and agencies track how AI search engines describe brands, compare competitors, and identify the sources shaping recommendations and create an action plan to increase AI visibility.
SEO, Analytics, AI
ZeroCostAI.pro1 clicks
No fake free trials. No credit card traps. 727 genuinely free AI tools tested and verified daily.
AI, Productivity, Developer tools
CloudPloy7 clicks
Add an API key and your agent deploys from Claude Code, Cursor, or any MCP client. Free forever: 1 small server, 1 app.
Developer tools, DevOps, AI
SkaleAgents4 clicks
Sign in with Google, publish specialist agents, connect MCP from your IDE. One control plane for AI DevOps work.
AI, DevOps, Developer tools
4 comments
Nora Callahan
Can I export replayable traces and benchmark results for offline analysis?
Hana Suzuki
How do I submit an agent to run against the available benchmarks and appear on the public leaderboard?
Marta Vidal
Can I compare two versions of an agent or harness on the same benchmark and inspect where their traces differ?
Felix Brandt
Can teams run private benchmark evaluations, or are all traces and results public?