reasoning benchmark
Tau2 Telecom evaluates conversational agents in shared telecommunications troubleshooting scenarios modeled as a Dec-POMDP.
Updated Sep 7, 2026
Higher score ranks better on this benchmark.
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAN | Score99.30% | Percentile100.00% | Participants36 | EvidenceC | Evaluated |
| Rank02 | ModelME | Score99.30% | Percentile97.14% | Participants36 | EvidenceC | Evaluated |
| Rank03 | ModelOP | Score98.90% | Percentile94.29% | Participants36 | EvidenceC | Evaluated |
| Rank04 | ModelOP | Score98.70% | Percentile91.43% | Participants36 | EvidenceC | Evaluated |
| Rank05 | ModelAN | Score98.20% | Percentile88.57% | Participants36 | EvidenceC | Evaluated |
| Rank06 | ModelOP | Score98.00% | Percentile85.71% | Participants36 | EvidenceC | Evaluated |
| Rank07 | ModelAN | Score97.90% | Percentile82.86% | Participants36 | EvidenceC | Evaluated |
| Rank08 | ModelXI | Score96.80% | Percentile80.00% | Participants36 | EvidenceC | Evaluated |
| Rank09 | ModelOP | Score96.70% | Percentile77.14% | Participants36 | EvidenceC | Evaluated |
| Rank10 | ModelOP | Score95.60% | Percentile74.29% | Participants36 | EvidenceC | Evaluated |
| Rank11 | ModelOP | Score95.60% | Percentile71.43% | Participants36 | EvidenceC | Evaluated |
| Rank12 | ModelOP | Score95.60% | Percentile68.57% | Participants36 | EvidenceC | Evaluated |
| Rank13 | ModelOP | Score93.40% | Percentile65.71% | Participants36 | EvidenceC | Evaluated |
| Rank14 | ModelAM | Score92.70% | Percentile62.86% | Participants36 | EvidenceC | Evaluated |
| Rank15 | ModelOP | Score92.50% | Percentile60.00% | Participants36 | EvidenceC | Evaluated |
| Rank16 | ModelME | Score91.50% | Percentile57.14% | Participants36 | EvidenceC | Evaluated |
| Rank17 | ModelMI | Score87.00% | Percentile54.29% | Participants36 | EvidenceC | Evaluated |
| Rank18 | ModelMI | Score87.00% | Percentile51.43% | Participants36 | EvidenceC | Evaluated |
| Rank19 | ModelCO | Score85.00% | Percentile48.57% | Participants36 | EvidenceC | Evaluated |
| Rank20 | ModelME | Score83.10% | Percentile45.71% | Participants36 | EvidenceC | Evaluated |
| Rank21 | ModelAN | Score83.00% | Percentile42.86% | Participants36 | EvidenceC | Evaluated |
| Rank22 | ModelAM | Score80.00% | Percentile40.00% | Participants36 | EvidenceC | Evaluated |
| Rank23 | ModelAM | Score76.00% | Percentile37.14% | Participants36 | EvidenceC | Evaluated |
| Rank24 | ModelME | Score73.68% | Percentile34.29% | Participants36 | EvidenceC | Evaluated |
| Rank25 | ModelLA | Score73.00% | Percentile31.43% | Participants36 | EvidenceC | Evaluated |
| Rank26 | ModelME | Score72.80% | Percentile28.57% | Participants36 | EvidenceC | Evaluated |
| Rank27 | ModelMI | Score71.70% | Percentile25.71% | Participants36 | EvidenceC | Evaluated |
| Rank28 | ModelMA | Score65.80% | Percentile22.86% | Participants36 | EvidenceC | Evaluated |
| Rank29 | ModelMA | Score65.80% | Percentile20.00% | Participants36 | EvidenceC | Evaluated |
| Rank30 | ModelNV | Score64.36% | Percentile17.14% | Participants36 | EvidenceC | Evaluated |
The leading models and scores on this benchmark.
A closer view of the leading scores on this benchmark.
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What Tau2 Telecom measures and how its scores work.
Tau2 Telecom is a telecom-domain benchmark in which agents and users use tools in a dual-control environment.
It measures Score as a ratio, testing coordination and communication capabilities.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
Common questions about Tau2 Telecom.
Claude Opus 4.6 is currently ranked first with 99.30%.
The current leaders are Claude Opus 4.6 (99.30%), LongCat-Flash-Thinking-2601 (99.30%), and GPT-5.4 (98.90%).
GPT-5.4 nano has the lowest matched official input price at $0.20 input / $1.3 output per 1M tokens.
The fastest matched records are GPT-5.5 (134.94 tok/s via OpenAI), GPT-4o (132.00 tok/s via OpenAI), and Claude Opus 4.6 (123.15 tok/s via Anthropic).
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.
It measures Score as a ratio, testing coordination and communication capabilities.
Yes. Higher values rank better for this benchmark.
36 model results are currently shown.
No. This benchmark is shown for reference but does not contribute to the overall score.
Ranking basisThis tau2 telecom AI model leaderboard uses descending score in the benchmark original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
Selection summary
Claude Opus 4.6 currently leads Tau2 Telecom with 99.30%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price, and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.