# Toolathlon Verified Pass³ leaderboard

Agentic 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/agentic-toolathlon-verified-pass3-all

Toolathlon Verified Pass cubed. Every model the catalog carries a published Toolathlon Verified Pass³ value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Agentic |
| Measure | All-three-trials pass rate |
| Tasks | 108 verified real-world tool-use tasks |
| Difficulty | Long-horizon application tool use |
| Published by | [Claude Opus 5 System Card](https://www-cdn.anthropic.com/c5fbac3f0b1280a933ebd26d3cb8bb9f5bdeaf48/Claude%20Opus%205%20System%20Card.pdf) |

Fraction of Toolathlon Verified tasks solved in all three independent trials.

## Toolathlon Verified Pass³ ranking

4 models with a published Toolathlon Verified Pass³ 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 | All-three-trials pass rate | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Claude Fable 5.1](https://modelscale.dev/models/claude-fable-5-1) | Anthropic | 73.1 | Observed 2026-11-21 · source benchlm:models |
| 1 | [Claude Opus 5](https://modelscale.dev/models/claude-opus-5) | Anthropic | 73.1 | Observed 2026-11-21 · source benchlm:models |
| 3 | [Claude Opus 5.5](https://modelscale.dev/models/claude-opus-5-5) | Anthropic | 72.2 | Observed 2026-11-21 · source benchlm:models |
| 4 | [Claude Sonnet 5.5](https://modelscale.dev/models/claude-sonnet-5-5) | Anthropic | 68.5 | Observed 2026-11-21 · 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.
