0.04
Confirmed task success · rank 13 of 46
- Unit
- IPS effect estimate, higher is better
- Range
- 0.02 to 0.05
- Sample
- 53688 observations
- Configuration
- GLM 5.2
- Measured
- 28 Sep 2026
- Not shown
- Not a success rate: the estimated effect of choosing this model as the agent's orchestrator, against an even mix of the models tested, inside Arena's own harness.
0.10
Praise over complaint · rank 6 of 46
- Unit
- IPS effect estimate, higher is better
- Range
- 0.07 to 0.12
- Sample
- 20071 observations
- Configuration
- GLM 5.2
- Measured
- 28 Sep 2026
- Not shown
- Not a success rate: the estimated effect of choosing this model as the agent's orchestrator, against an even mix of the models tested, inside Arena's own harness.
0.04
Steerability · rank 6 of 46
- Unit
- IPS effect estimate, higher is better
- Range
- 0.03 to 0.06
- Sample
- 67812 observations
- Configuration
- GLM 5.2
- Measured
- 28 Sep 2026
- Not shown
- Not a success rate: the estimated effect of choosing this model as the agent's orchestrator, against an even mix of the models tested, inside Arena's own harness.
0.00
Tool grounding · rank 1 of 46
- Unit
- IPS effect estimate, higher is better
- Range
- 0.00 to 0.00
- Sample
- 5594512 observations
- Configuration
- GLM 5.2
- Measured
- 28 Sep 2026
- Not shown
- Not a success rate: the estimated effect of choosing this model as the agent's orchestrator, against an even mix of the models tested, inside Arena's own harness.
1,463
Overall · rank 26 of 177
- Unit
- Arena rating, higher is better
- Range
- 1,459 to 1,466
- Sample
- 43341 votes
- Configuration
- GLM 5.2 (max)
- Measured
- 25 Sep 2026
- Not shown
- Not an error rate: claims that cannot be checked on the web are skipped, and preference still carries most of the weight.
1,463
Business, management and finance · rank 26 of 402
- Unit
- Arena rating, higher is better
- Range
- 1,456 to 1,471
- Sample
- 8496 votes
- Configuration
- GLM 5.2 (max)
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,455
Creative writing · rank 13 of 407
- Unit
- Arena rating, higher is better
- Range
- 1,447 to 1,462
- Sample
- 8448 votes
- Configuration
- GLM 5.2 (max)
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,500
Expert prompts · rank 14 of 359
- Unit
- Arena rating, higher is better
- Range
- 1,491 to 1,509
- Sample
- 4697 votes
- Configuration
- GLM 5.2 (max)
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,470
Instruction following · rank 16 of 409
- Unit
- Arena rating, higher is better
- Range
- 1,464 to 1,476
- Sample
- 15327 votes
- Configuration
- GLM 5.2 (max)
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,476
Overall · rank 19 of 409
- Unit
- Arena rating, higher is better
- Range
- 1,471 to 1,480
- Sample
- 43570 votes
- Configuration
- GLM 5.2 (max)
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,459
Writing, literature and language · rank 18 of 408
- Unit
- Arena rating, higher is better
- Range
- 1,452 to 1,466
- Sample
- 11517 votes
- Configuration
- GLM 5.2 (max)
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
13.4%
Consistency (pass^4) · rank 13 of 21
- Unit
- % of tasks, higher is better
- Configuration
- glm-5.2
- Measured
- 24 Jul 2026
- Not shown
- Not your policies or customers: the tasks and policies are tau2's own, and the customer is simulated by gpt-5.2.
37.1%
Task success (pass^1) · rank 11 of 21
- Unit
- % of tasks, higher is better
- Configuration
- glm-5.2
- Measured
- 24 Jul 2026
- Not shown
- Not your policies or customers: the tasks and policies are tau2's own, and the customer is simulated by gpt-5.2.
12.0%
Requested-length error · rank 98 of 370
- Unit
- % off the requested word count, lower is better
- Configuration
- glm-5.2
- Measured
- 21 Jun 2026
- Not shown
- Not other format limits such as character counts or bullet counts.
18.0%
Requested-length error · rank 150 of 370
- Unit
- % off the requested word count, lower is better
- Configuration
- glm-5.2
- Measured
- 21 Jun 2026
- Not shown
- Not other format limits such as character counts or bullet counts.
0.37
Style adherence · rank 81 of 370
- Unit
- score from 0 to 1, higher is better
- Configuration
- glm-5.2
- Measured
- 21 Jun 2026
- Not shown
- Not brand voice on your own examples: UGI's prompts are private and lean towards creative writing.
0.35
Style adherence · rank 190 of 370
- Unit
- score from 0 to 1, higher is better
- Configuration
- glm-5.2
- Measured
- 21 Jun 2026
- Not shown
- Not brand voice on your own examples: UGI's prompts are private and lean towards creative writing.
61.2
Writing score · rank 92 of 370
- Unit
- score out of 100, higher is better
- Configuration
- glm-5.2
- Measured
- 21 Jun 2026
- Not shown
- Not business copy quality: UGI's prompts are private and lean towards creative writing, and models that often refuse get no score.
67.0
Writing score · rank 47 of 370
- Unit
- score out of 100, higher is better
- Configuration
- glm-5.2
- Measured
- 21 Jun 2026
- Not shown
- Not business copy quality: UGI's prompts are private and lean towards creative writing, and models that often refuse get no score.
