reasoning benchmark
API-Bank
A comprehensive benchmark for tool-augmented LLMs that evaluates API planning, retrieval, and calling capabilities. Contains 314 tool-use dialogues with 753 API calls across 73 API tools, designed to assess how effectively LLMs can utilize external tools and overcome obstacles in tool leveraging.
Updated Aug 7, 2026
Published models3
Registry coverage3
MetricScore
EvidenceB
API-Bank leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | ME | 92.0% | 100.0% | 3 | C | |
| 2 | ME | 90.0% | 50.0% | 3 | C | |
| 3 | ME | 82.6% | 0.0% | 3 | C |
Score distribution
Top published rows on the benchmark's original scale.
API-Bank highlights
The top published results on this benchmark's own scale.
What is API-Bank?
Definition and scoring fields from the benchmark registry.
A comprehensive benchmark for tool-augmented LLMs that evaluates API planning, retrieval, and calling capabilities. Contains 314 tool-use dialogues with 753 API calls across 73 API tools, designed to assess how effectively LLMs can utilize external tools and overcome obstacles in tool leveraging.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- API-Bank
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- api-bank|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about API-Bank.
Which model scores highest on API-Bank?
Llama 3.1 405B Instruct is currently ranked first with 92.0%.
What does API-Bank measure?
A comprehensive benchmark for tool-augmented LLMs that evaluates API planning, retrieval, and calling capabilities. Contains 314 tool-use dialogues with 753 API calls across 73 API tools, designed to assess how effectively LLMs can utilize external tools and overcome obstacles in tool leveraging.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
3 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.