Benchmarks / Microsoft STATE-Bench

Reported by Microsoft STATE-Bench

Microsoft STATE-Bench

Mean rating of the conversation from the simulated user's side.

Data older than 90 days: last updated 1 Jul 2026

Results dated
1 Jul 2026
Models
2
Unit
score from 1 to 5
Licence
MIT License

Microsoft STATE-Bench in brief

What it measures. Microsoft STATE-Bench is a third-party benchmark, shown here with credit. Measured here: Mean rating of the conversation from the simulated user's side.

Latest results. , 2 models ranked.

Top 2 on User experience (score from 1 to 5, higher is better):

  1. GPT-5.4 (high reasoning) (OpenAI): 3.67
  2. GPT-5.5 (high reasoning) (OpenAI): 3.62

How to cite. Microsoft STATE-Bench, via Spring Prompt, results dated 1 July 2026. https://springprompt.com/benchmarks/statebench These results belong to Microsoft STATE-Bench, not to us. Licence: MIT License. Credit Microsoft STATE-Bench first.

User experience:

2 results · score from 1 to 5, higher is better. Choose a model to highlight it.Clear highlight

  1. 1 GPT-5.4 (high reasoning)OpenAI 3.67
  2. 2 GPT-5.5 (high reasoning)OpenAI 3.62

Full results

STATE-Bench v0.8: user experience, score from 1 to 5, higher is better
#ModelUser experience
score from 1 to 5, higher is better
Price
$ per million tokens, in / out
1 GPT-5.4 (high reasoning)OpenAI
3.67
$2.50 / $15
2 GPT-5.5 (high reasoning)OpenAI
3.62
$5 / $30

Swipe the table sideways for more columns.

Source: Microsoft STATE-Bench, via Spring Prompt, springprompt.com/benchmarks/statebench. Test your own prompt on these models →

Ranks follow the score as shown, so equal numbers share a rank. Results as published by Microsoft STATE-Bench; we do not re-run them.

What it measures

Mean rating of the conversation from the simulated user's side.

What it does not measure

Not real customer satisfaction: the ratings come from a model-based judge.

4 results from Microsoft STATE-Bench not ranked here · show why

We rank a result only when we can tie it to a specific model you can use. These are left out:

  • Agent systems with memory, not models on their own: 4

STATE-Bench by Microsoft and STATE-Bench contributors, MIT License.