Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Loading page…
Preferred languages
More languages supported by Google Translate
Open the selector to load the current language catalog.
Translations are provided by Google Translate.
Agentic benchmark
Every model the catalog carries a published ToolSandbox value for, ranked by that value.
Apple benchmark of stateful, conversational tool use, where tool calls change state and a simulated user takes part. 3 models in the ModelChoice catalog have a published value.
Each point is a model with a published ToolSandbox value and complete token prices, on the 75% input / 25% output blend. Frontier models are spaced evenly along the cost axis.
3 models with a published ToolSandbox value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.
| Rank | Model | Provider | Average accuracy |
|---|---|---|---|
| 1 | LiquidAI | 80.36 | |
| 2 | NVIDIA | 74.13 | |
| 3 | NVIDIA | 68.77 |
What the benchmark measures and how its value is published, from the source’s own record.
ToolSandbox is an agentic benchmark published by ToolSandbox: A Stateful, Conversational, Interactive Evaluation Benchmark for LLM Tool Use Capabilities. Apple benchmark of stateful, conversational tool use, where tool calls change state and a simulated user takes part. Its evaluation set is described by the source as stateful tool-use conversations. The source rates its difficulty as tool use with changing state.
Each model's result is published by ToolSandbox: A Stateful, Conversational, Interactive Evaluation Benchmark for LLM Tool Use Capabilities as average accuracy. ModelChoice prints that value to 2 decimal places, exactly as published, and never converts or rescales it. This page ranks the 3 models that carry a published value, highest value first; a model the source never scored is left out rather than ranked at zero. The source does not state whether a higher value is the better result, so neither does this page.
Evidence key: Observed
Rows are ordered by the value ToolSandbox: A Stateful, Conversational, Interactive Evaluation Benchmark for LLM Tool Use Capabilities published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.