# τ²-bench

> τ²-bench results for 6 AI models we list, run by Sierra, with each result's setting, date and source.

- Canonical: https://www.anchorterminal.com/benchmarks/tau2-bench/
- Markdown: https://www.anchorterminal.com/benchmarks/tau2-bench/index.md (~1,750 tokens)
- Slim: https://www.anchorterminal.com/benchmarks/tau2-bench/index.min.md (~1,630 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/benchmarks/tau2-bench/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-21

A model working as a customer-service agent under a written policy, with tools and a simulated customer, in one domain at a time (airline, retail, telecom, banking), in text or by voice. Each domain is scored on its own.

- Score: % of tasks passed on the first try (pass^1), higher is better
- Run by: Sierra
- Status: Current, newest result 29 October 2026
- Updated: as submissions land
- Data: https://github.com/sierra-research/tau2-bench/tree/main/web/leaderboard/public/submissions (MIT, https://github.com/sierra-research/tau2-bench/blob/main/LICENSE), read 2026-11-20
- Method: https://taubench.com/

## Results: text: banking knowledge

4 models we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [Kimi K3](https://www.anchorterminal.com/tools/kimi-k3.md) | 37.1% |  | Independent | effort max, retrieval alltools, submission type standard, tau2 version 1.0.1, user simulator gpt-5.2 | 2026-08-04 | https://taubench.com/ |
| 2 | [Gemini 3.1 Pro (preview)](https://www.anchorterminal.com/tools/vertex-ai.md) | 26.0% |  | Independent | effort high, retrieval alltools, submission type standard, tau2 version 1.0.1, user simulator gpt-5.2 | 2026-05-05 | https://taubench.com/ |
| 3 | [Inkling](https://www.anchorterminal.com/tools/inkling.md) | 25.0% |  | Independent | effort max, retrieval alltools, submission type standard, tau2 version 1.0.1, user simulator gpt-5.2 | 2026-08-04 | https://taubench.com/ |
| 4 | [Qwen3.5-397B-A17B](https://www.anchorterminal.com/tools/qwen3-5-397b-a17b.md) | 9.8% |  | Independent | effort enabled, retrieval text-emb-3-large, submission type standard, tau2 version 1.0.1, user simulator gpt-5.2 | 2026-03-02 | https://taubench.com/ |

## Results: voice: banking knowledge

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [GPT-Live](https://www.anchorterminal.com/tools/gpt-live.md) | 37.1% |  | Submitted by the maker | effort Backend: GPT-6 Astra (high), retrieval alltools, submission type custom, tau2 version 1.0.1, user simulator gpt-5.5 (xhigh) | 2026-09-16 | https://taubench.com/ |

## Results: voice-v2-entity

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [GPT-Live](https://www.anchorterminal.com/tools/gpt-live.md) | 60.3% |  | Independent | agent provider openai_live, submission type standard | 2026-10-20 | https://taubench.com/ |

## Results: voice-v2-banking

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [GPT-Live](https://www.anchorterminal.com/tools/gpt-live.md) | 33.2% |  | Independent | agent provider openai_live, submission type standard | 2026-10-20 | https://taubench.com/ |

## Results: voice-v2

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [GPT-Live](https://www.anchorterminal.com/tools/gpt-live.md) | 79.9% |  | Independent | agent provider openai_live, submission type standard | 2026-10-20 | https://taubench.com/ |

## Results: text: telecom (legacy)

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [DeepSeek-V3.2](https://www.anchorterminal.com/tools/deepseek-v3-2.md) | 96.2% |  | Submitted by the maker | tau2 version v0.1.3, user simulator DeepSeek-V3.2 | 2025-12-01 | https://taubench.com/ |

## Results: text: telecom

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [Qwen3.5-397B-A17B](https://www.anchorterminal.com/tools/qwen3-5-397b-a17b.md) | 97.8% |  | Independent | effort enabled, submission type standard, tau2 version 1.0.1, user simulator gpt-5.2 | 2026-03-02 | https://taubench.com/ |

## Results: text: retail (legacy)

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [DeepSeek-V3.2](https://www.anchorterminal.com/tools/deepseek-v3-2.md) | 81.1% |  | Submitted by the maker | tau2 version v0.1.3, user simulator DeepSeek-V3.2 | 2025-12-01 | https://taubench.com/ |

## Results: text: retail

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [Qwen3.5-397B-A17B](https://www.anchorterminal.com/tools/qwen3-5-397b-a17b.md) | 84.4% |  | Independent | effort enabled, submission type standard, tau2 version 1.0.1, user simulator gpt-5.2 | 2026-03-02 | https://taubench.com/ |

## Results: text: airline (legacy)

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [DeepSeek-V3.2](https://www.anchorterminal.com/tools/deepseek-v3-2.md) | 63.8% |  | Submitted by the maker | tau2 version v0.1.3, user simulator DeepSeek-V3.2 | 2025-12-01 | https://taubench.com/ |

## Results: text: airline

1 model we list, each at its best setting, best first.

| # | Model | Score | Range | Who ran it | Setting | Date | Source |
| --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | [Qwen3.5-397B-A17B](https://www.anchorterminal.com/tools/qwen3-5-397b-a17b.md) | 81.5% |  | Independent | effort enabled, submission type standard, tau2 version 1.0.1, user simulator gpt-5.2 | 2026-03-02 | https://taubench.com/ |

A rank is among the models we list with a result, not a place on the whole leaderboard, and a model offered by several hosts is shown once, on its own listing. These results don't move our grades.

Data: τ²-bench, Sierra Research, MIT, MIT. All benchmarks: https://www.anchorterminal.com/benchmarks/index.md

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
