agents benchmark
Kimi Claw 24/7 Bench
Kimi Claw 24/7 Bench is Moonshot AI's in-house benchmark for evaluating long-horizon agentic performance in persistent, multi-day coworking tasks. It spans 17 professional scenarios across 610 evaluation points, covering software engineering, ML research, recruiting, trading, and marketing tasks executed through the OpenClaw harness.
Updated Aug 11, 2026
Models1
Model coverage1
MetricScore
EvidenceB
Kimi Claw 24/7 Bench Ranking
Higher score ranks better on this benchmark.
| 1 | MA | 46.9% | 100.0% | 1 | C |
Kimi Claw 24/7 Bench Highlights
The leading models and scores on this benchmark.
What is Kimi Claw 24/7 Bench?
What Kimi Claw 24/7 Bench measures and how its scores work.
Kimi Claw 24/7 Bench is Moonshot AI's in-house benchmark for evaluating long-horizon agentic performance in persistent, multi-day coworking tasks. It spans 17 professional scenarios across 610 evaluation points, covering software engineering, ML research, recruiting, trading, and marketing tasks executed through the OpenClaw harness.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- Kimi Claw 24/7 Bench
- Modality
- text
- Primary category
- agents
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- kimi-claw-24-7-bench|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about Kimi Claw 24/7 Bench.
Which model scores highest on Kimi Claw 24/7 Bench?
Kimi K2.7 Code is currently ranked first with 46.9%.
What does Kimi Claw 24/7 Bench measure?
Kimi Claw 24/7 Bench is Moonshot AI's in-house benchmark for evaluating long-horizon agentic performance in persistent, multi-day coworking tasks. It spans 17 professional scenarios across 610 evaluation points, covering software engineering, ML research, recruiting, trading, and marketing tasks executed through the OpenClaw harness.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 model results are currently shown.
Does this benchmark affect the overall score?
No. This benchmark is shown for reference but does not contribute to the overall score.