Agents / Benchmark profile
τ³-bench is the current evolution of Sierra's tool-agent-user framework, adding corrected task releases and newer knowledge and voice evaluation modes alongside airline, retail, and telecom.
Observed results
10 ranked models · higher is better · labels show rank and score
| Model | Rank | Model ID | Provider | Score | Unit |
|---|---|---|---|---|---|
| Mistral Medium 3.5 128B | #1 | mistral-medium-3-5-128b | Mistral AI | 91.4 | percent |
| MiMo-V2.5-Pro | #2 | mimo-v2-5-pro | Xiaomi | 72.9 | percent |
| Nemotron 3 Ultra | #3 | nemotron-3-ultra | NVIDIA | 70.9 | percent |
| Qwen3.6 Plus | #4 | qwen3-6-plus | Alibaba Cloud | 70.7 | percent |
| GLM-5.1 | #5 | glm-5-1 | Z.AI | 70.6 | percent |
| Claude Opus 4.5 | #6 | claude-opus-4-5 | Anthropic | 70.2 | percent |
| Qwen3.5 397B A17B | #7 | qwen3-5-397b | Alibaba Cloud | 68.4 | percent |
| Qwen3.6-35B-A3B | #8 | qwen3-6-35b-a3b | Alibaba Cloud | 67.2 | percent |
| Kimi K2.5 | #9 | kimi-k2-5 | Moonshot AI | 65.7 | percent |
| GLM-5 | #10 | glm-5 | Z.AI | 65.6 | percent |
One best compatible score per canonical product · higher is better
| Rank | Model | Weights | Evidence | Score |
|---|---|---|---|---|
| 01 | Mistral Medium 3.5 128BMistral AI | Open weights | Source carrier | 91.4% |
| 02 | MiMo-V2.5-ProXiaomi | Open weights | Source carrier | 72.9% |
| 03 | Nemotron 3 UltraNVIDIA | Open weights | Source carrier | 70.9% |
| 04 | Qwen3.6 PlusAlibaba Cloud | Closed weights | Source carrier | 70.7% |
| 05 | GLM-5.1Z.AI | Open weights | Source carrier | 70.6% |
| 06 | Claude Opus 4.5Anthropic | Closed weights | Source carrier | 70.2% |
| 07 | Qwen3.5 397B A17BAlibaba Cloud | Open weights | Source carrier | 68.4% |
| 08 | Qwen3.6-35B-A3BAlibaba Cloud | Open weights | Source carrier | 67.2% |
| 09 | Kimi K2.5Moonshot AI | Open weights | Source carrier | 65.7% |
| 10 | GLM-5Z.AI | Open weights | Source carrier | 65.6% |
Each distinct published measurement is retained. Same-snapshot canonical and configuration projections appear once.
| Published model / configuration | Score | Evidence & protocol | Source & dates |
|---|---|---|---|
Exact BenchLM registry variant Mistral Medium 3.5 128B; bulk export does not retain a complete upstream harness configuration.Mistral Medium 3.5 128BExact identitySource label without a registered configuration ID Canonical product: mistral-medium-3-5-128b | 91.4% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
Exact BenchLM registry variant MiMo-V2.5-Pro; bulk export does not retain a complete upstream harness configuration.MiMo-V2.5-ProExact identitySource label without a registered configuration ID Canonical product: mimo-v2-5-pro | 72.9% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
Exact BenchLM registry variant Nemotron 3 Ultra; bulk export does not retain a complete upstream harness configuration.Nemotron 3 UltraExact identitySource label without a registered configuration ID Canonical product: nemotron-3-ultra | 70.9% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
Exact BenchLM registry variant Qwen3.6 Plus; bulk export does not retain a complete upstream harness configuration.Qwen3.6 PlusExact identitySource label without a registered configuration ID Canonical product: qwen3-6-plus | 70.7% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
Exact BenchLM registry variant GLM-5.1; bulk export does not retain a complete upstream harness configuration.GLM-5.1Exact identitySource label without a registered configuration ID Canonical product: glm-5-1 | 70.6% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
Exact BenchLM registry variant Claude Opus 4.5; bulk export does not retain a complete upstream harness configuration.Claude Opus 4.5Exact identitySource label without a registered configuration ID Canonical product: claude-opus-4-5 | 70.2% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
Exact BenchLM registry variant Qwen3.5 397B; bulk export does not retain a complete upstream harness configuration.Qwen3.5 397B A17BExact identitySource label without a registered configuration ID Canonical product: qwen3-5-397b | 68.4% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
Exact BenchLM registry variant Qwen3.6-35B-A3B; bulk export does not retain a complete upstream harness configuration.Qwen3.6-35B-A3BExact identitySource label without a registered configuration ID Canonical product: qwen3-6-35b-a3b | 67.2% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
Exact BenchLM registry variant Kimi K2.5; bulk export does not retain a complete upstream harness configuration.Kimi K2.5Exact identitySource label without a registered configuration ID Canonical product: kimi-k2-5 | 65.7% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
Exact BenchLM registry variant GLM-5; bulk export does not retain a complete upstream harness configuration.GLM-5Exact identitySource label without a registered configuration ID Canonical product: glm-5 | 65.6% | Source carriersource-checkedVersion & system2026 Source-native system | BenchLM public datasets — 2026-08-01Observed Checked |
From result to context
τ³-bench is the current evolution of Sierra's tool-agent-user framework, adding corrected task releases and newer knowledge and voice evaluation modes alongside airline, retail, and telecom.
The summary shows one best compatible source result per canonical product. Versions, effort settings and execution systems remain attached to the underlying records.
Scores from different versions or harnesses may not be interchangeable. The published source rows preserve those distinctions and their original units. A source result is not automatically an input to a current index.
This catalogue entry preserves available source evidence. Current indices admit only their specifically reviewed tracks and configurations.
Contamination risk: Unknown. Lifecycle: Active.
