BlogLLM Benchmark Scores as Published, Side by Side
HackerBench: Results as Published
Every result the catalog holds for HackerBench: 4 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 CybersecurityMaintainer xAIData version 2026103001, as of 2026-10-30
The xAI Grok 4.7 launch page calls HackerBench v0.3 its own benchmark for risky and malicious cyber tasks. The reported result is the share of risky dual-use prompts that the model let through, and the page says Grok 4.7 let through 3.3 percent while showing the highest safety of the models compared. The page does not explain how prompts are built or how each answer is judged.
HackerBench v0.3 benign refusal
lower is better
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPT-5.6 SolOpenAI, released 2026-07-09 | 0% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: max |
| Grok 4.7xAI, released 2026-09-21 | 0.31% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: high |
| 0% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: xhigh | |
| Grok 4.6xAI, released 2026-08-12 | 0% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: high |
| Grok 4.5xAI, released 2026-07-16 | 0% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: high |
HackerBench v0.3 harmful and dual-use compliance
lower is better
| Model | Value | Printed by | Setup as printed |
|---|---|---|---|
| GPT-5.6 SolOpenAI, released 2026-07-09 | 35.7% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: max |
| Grok 4.7xAI, released 2026-09-21 | 4.02% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: xhigh |
| 3.31% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: high | |
| Grok 4.6xAI, released 2026-08-12 | 5.93% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: high |
| Grok 4.5xAI, released 2026-07-16 | 7.8% | xAI, Grok 4.7 model cardchecked 2026-10-14 | Effort: high |
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.