1,452
Overall · rank 46 of 177
- Unit
- Arena rating, higher is better
- Range
- 1,448 to 1,456
- Sample
- 28919 votes
- Configuration
- GLM 5
- 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.
1,451
Business, management and finance · rank 42 of 402
- Unit
- Arena rating, higher is better
- Range
- 1,443 to 1,459
- Sample
- 5890 votes
- Configuration
- GLM 5
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,447
Creative writing · rank 20 of 407
- Unit
- Arena rating, higher is better
- Range
- 1,438 to 1,456
- Sample
- 4799 votes
- Configuration
- GLM 5
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,481
Expert prompts · rank 33 of 359
- Unit
- Arena rating, higher is better
- Range
- 1,469 to 1,493
- Sample
- 2597 votes
- Configuration
- GLM 5
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,446
Instruction following · rank 47 of 409
- Unit
- Arena rating, higher is better
- Range
- 1,439 to 1,452
- Sample
- 9562 votes
- Configuration
- GLM 5
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,457
Overall · rank 48 of 409
- Unit
- Arena rating, higher is better
- Range
- 1,453 to 1,461
- Sample
- 29152 votes
- Configuration
- GLM 5
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
1,447
Writing, literature and language · rank 28 of 408
- Unit
- Arena rating, higher is better
- Range
- 1,440 to 1,455
- Sample
- 6972 votes
- Configuration
- GLM 5
- Measured
- 25 Sep 2026
- Not shown
- Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.
49.4%
Average · rank 17 of 34
- Unit
- % resolved, higher is better
- Configuration
- glm-5
- Measured
- 30 Jun 2026
- Not shown
- Not the model alone: the score is for the model inside the OpenHands agent, at the SDK version shown.
35.3%
Front end (SWE-Bench Multimodal) · rank 17 of 34
- Unit
- % resolved, higher is better
- Configuration
- glm-5
- Measured
- 30 Jun 2026
- Not shown
- Not the model alone: the score is for the model inside the OpenHands agent, at the SDK version shown.
31.2%
Greenfield (Commit0) · rank 13 of 34
- Unit
- % resolved, higher is better
- Configuration
- glm-5
- Measured
- 30 Jun 2026
- Not shown
- Not the model alone: the score is for the model inside the OpenHands agent, at the SDK version shown.
60.0%
Information gathering (GAIA) · rank 18 of 34
- Unit
- % resolved, higher is better
- Configuration
- glm-5
- Measured
- 30 Jun 2026
- Not shown
- Not the model alone: the score is for the model inside the OpenHands agent, at the SDK version shown.
73.4%
Issue resolution (SWE-Bench) · rank 20 of 34
- Unit
- % resolved, higher is better
- Configuration
- glm-5
- Measured
- 30 Jun 2026
- Not shown
- Not the model alone: the score is for the model inside the OpenHands agent, at the SDK version shown.
47.3%
Testing (SWT-Bench) · rank 27 of 34
- Unit
- % resolved, higher is better
- Configuration
- glm-5
- Measured
- 30 Jun 2026
- Not shown
- Not the model alone: the score is for the model inside the OpenHands agent, at the SDK version shown.
70.0%
Consistency (pass^4) · rank 2 of 8
- Unit
- % of tasks, higher is better
- Configuration
- glm-5
- Measured
- 27 Feb 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.
82.5%
Task success (pass^1) · rank 3 of 8
- Unit
- % of tasks, higher is better
- Configuration
- glm-5
- Measured
- 27 Feb 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.
3.1%
Consistency (pass^4) · rank 2 of 2
- Unit
- % of tasks, higher is better
- Configuration
- glm-5
- Measured
- 27 Feb 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.
9.8%
Task success (pass^1) · rank 1 of 2
- Unit
- % of tasks, higher is better
- Configuration
- glm-5
- Measured
- 27 Feb 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.
43.9%
Consistency (pass^4) · rank 7 of 8
- Unit
- % of tasks, higher is better
- Configuration
- glm-5
- Measured
- 27 Feb 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.
73.7%
Task success (pass^1) · rank 7 of 8
- Unit
- % of tasks, higher is better
- Configuration
- glm-5
- Measured
- 27 Feb 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.
62.3%
Consistency (pass^4) · rank 7 of 8
- Unit
- % of tasks, higher is better
- Configuration
- glm-5
- Measured
- 27 Feb 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.
86.8%
Task success (pass^1) · rank 6 of 8
- Unit
- % of tasks, higher is better
- Configuration
- glm-5
- Measured
- 27 Feb 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.
36.0%
Requested-length error · rank 281 of 370
- Unit
- % off the requested word count, lower is better
- Configuration
- glm-5
- Measured
- 11 Feb 2026
- Not shown
- Not other format limits such as character counts or bullet counts.
34.0%
Requested-length error · rank 269 of 370
- Unit
- % off the requested word count, lower is better
- Configuration
- glm-5
- Measured
- 11 Feb 2026
- Not shown
- Not other format limits such as character counts or bullet counts.
0.40
Style adherence · rank 23 of 370
- Unit
- score from 0 to 1, higher is better
- Configuration
- glm-5
- Measured
- 11 Feb 2026
- Not shown
- Not brand voice on your own examples: UGI's prompts are private and lean towards creative writing.
0.38
Style adherence · rank 58 of 370
- Unit
- score from 0 to 1, higher is better
- Configuration
- glm-5
- Measured
- 11 Feb 2026
- Not shown
- Not brand voice on your own examples: UGI's prompts are private and lean towards creative writing.
49.2
Writing score · rank 161 of 370
- Unit
- score out of 100, higher is better
- Configuration
- glm-5
- Measured
- 11 Feb 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.
55.0
Writing score · rank 127 of 370
- Unit
- score out of 100, higher is better
- Configuration
- glm-5
- Measured
- 11 Feb 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.
99.7%
Answer rate · rank 44 of 108
- Unit
- % of documents, higher is better
- Configuration
- GLM 5
- Measured
- 22 Sep 2026
- Not shown
- Not a quality score: a low rate usually means content filters were triggered, and hallucination rates are measured on answered documents only.
10.1%
Hallucination rate · rank 59 of 108
- Unit
- % of summaries, lower is better
- Configuration
- GLM 5
- Measured
- 22 Sep 2026
- Not shown
- Not errors in open questions or other tasks: only summarisation, judged by Vectara's own model (HHEM-2.3), not by people, on news-style documents rather than your data.
