BlogLLM Benchmark Scores as Published, Side by Side
OpenAI Internal Computer Use Safety: Results as Published
Every result the catalog holds for OpenAI Internal Computer Use Safety: 5 models from 2 developers, read from 1 page. Each number is as printed by the company named beside it; nothing is ranked or rescored. Models are listed by developer, newest first.
Area Safety and AlignmentMaintainer OpenAIIntroduced 2026Data version 2026103001, as of 2026-10-30
The GPT-6 Astra system card's section 'Avoiding Misaligned Behavior in Realistic Work Environments' tests computer- and browser-use agents on realistic tasks and reports the share of runs with a misaligned outcome, such as an unauthorized transaction or data exfiltration. Lower is better. The printed name, with and without the AutoReview variant, is not found in the card text read here, so the mapping is a judgment.
Internal computer use safety benchmark
lower is better
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPT-6 AstraOpenAI, released 2026-09-03 | 2.4% | OpenAI, GPT-6 Astrachecked 2026-10-14 | - |
| GPT-5.6 SolOpenAI, released 2026-07-09 | 22% | OpenAI, GPT-6 Astrachecked 2026-10-14 | - |
| Claude Fable 5.1Anthropic, released 2026-09-01 | 9.5% | OpenAI, GPT-6 Astrachecked 2026-10-14 | - |
| Claude Opus 5Anthropic, released 2026-07-24 | 11.5% | OpenAI, GPT-6 Astrachecked 2026-10-14 | - |
| Claude Fable 5Anthropic, released 2026-06-09 | 18.3% | OpenAI, GPT-6 Astrachecked 2026-10-14 | - |
Internal computer use safety benchmark (with AutoReview)
lower is better
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPT-6 AstraOpenAI, released 2026-09-03 | 1.8% | OpenAI, GPT-6 Astrachecked 2026-10-14 | - |
| GPT-5.6 SolOpenAI, released 2026-07-09 | 4.3% | OpenAI, GPT-6 Astrachecked 2026-10-14 | - |
Every number here is read from the same catalog file as the benchmark table, data version 2026103001, and is the number the table shows for that model and test. Where one page printed a model under several setups, each value is listed. The catalog is public in armanobosyan/sugra-llm-benchmarks; a wrong number can be reported from the table, with Report an error under the value, or as an issue in that repository.