# When2Call (Decision Index) leaderboard

external benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelscale.dev/leaderboards/benchmarks/external-aa-decision-when2-call

Every model the catalog carries a published When2Call (Decision Index) value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | external |
| Measure | Accuracy |
| Tasks | — |
| Difficulty | Tool-use decisions |
| Published by | [Artificial Analysis Decision Index v1.0](https://artificialanalysis.ai/articles/artificial-analysis-decision-index) |

When2Call is one component of the Artificial Analysis Decision Index v1.0, run by Artificial Analysis. It measures accuracy on NVIDIA's tool-calling evaluation: given a conversation and tools, choose to answer directly, call a tool, ask for more information, or decline.

## When2Call (Decision Index) ranking

13 models with a published When2Call (Decision Index) 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 | Accuracy | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [elephant-1](https://modelscale.dev/models/elephant-1) | Baseten | 94.8 | Observed 2026-10-23 · source benchlm:models |
| 2 | [d1-ultra](https://modelscale.dev/models/d1-ultra) | Liquid AI | 92.6 | Observed 2026-10-23 · source benchlm:models |
| 3 | [Perplexity Decider v1.2 27B](https://modelscale.dev/models/pplx-decider-v1-2) | Perplexity | 87.7 | Observed 2026-10-23 · source benchlm:models |
| 4 | [Celeris-1 Decision](https://modelscale.dev/models/celeris-1-decision) | Celeris | 83.7 | Observed 2026-10-23 · source benchlm:models |
| 4 | [GLiDE](https://modelscale.dev/models/fastino-glide) | Fastino | 83.7 | Observed 2026-10-23 · source benchlm:models |
| 6 | [Microsoft-Decision-1](https://modelscale.dev/models/microsoft-decision-1) | Microsoft | 81.9 | Observed 2026-10-23 · source benchlm:models |
| 7 | [Jev 1.14](https://modelscale.dev/models/jev-1-14) | TypeSafe AI | 80.8 | Observed 2026-10-23 · source benchlm:models |
| 8 | [Mercury Decide](https://modelscale.dev/models/mercury-decide) | Inception | 79.3 | Observed 2026-10-23 · source benchlm:models |
| 9 | [d1](https://modelscale.dev/models/liquid-d1) | Liquid AI | 77.4 | Observed 2026-10-23 · source benchlm:models |
| 10 | [Perplexity Decider Flash v1.2 4B](https://modelscale.dev/models/pplx-decider-flash-v1-2) | Perplexity | 74.4 | Observed 2026-10-23 · source benchlm:models |
| 11 | [Clef](https://modelscale.dev/models/cloudflare-clef) | Cloudflare | 71.0 | Observed 2026-10-23 · source benchlm:models |
| 12 | [Clef-Flash](https://modelscale.dev/models/cloudflare-clef-flash) | Cloudflare | 64.4 | Observed 2026-10-23 · source benchlm:models |
| 13 | [Clef-Omni](https://modelscale.dev/models/cloudflare-clef-omni) | Cloudflare | 62.1 | Observed 2026-10-23 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.