WebTailBench-Refusals
Fara1.5-27B 100% Computer-use agents receive 111 harmful browser tasks in seven policy categories, such as illegal activity and misinformation. Scored as the share of tasks safely refused.
Released with Fara-7B as WebTailBench-Refusals.tsv in the WebTailBench dataset. The webtailbench record lists this name as an alias, which should move here.
- Made by
- Microsoft
- Released
- November 2025
- Kind
- Refusing harmful requests
- Level
- Everyday
- Scored
- Graded by a model
- Answers
- Agent task
- Tasks
- All public
- Last in a system card
- June 2026
- Last model added
- June 2026
Fara1.5-27B leads with 100%
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
- 1 Fara1.5-27B 100%
- 2 Fara-7B 81.9%
Source: Labs’ model cards
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | Fara1.5-27B | 100% | Jun 2026 | Lab model card | |
| 2 | Fara-7B | 81.9% | Nov 2025 | Lab model card |
Microsoft’s own number for Fara1.5-27B passed 85% in June 2026
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Labs’ model cards
Labs report it only in system cards
We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark. Some system cards do, below.
2 system cards report it, from 1 lab
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| Jun 18, 2026 | Fara1.5-27B technical report | 100 | WebTailBench-Refusals, 111 tasks, the whole Fara1.5 family refuses 100% (Section 4.3 text) | Safety |
| Nov 24, 2025 | Fara-7B technical report | 81.9 | WebTailBench-Refusals, harmful tasks safely refused (Table 14) | Safety |
More in Safety and alignment
- FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
- StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is.
Inkling 98.6% - Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts.
Gemini 4 Carbon 0% - DistressBench In use Models reply as a chatbot to 718 simulated users in psychological crisis, in single-turn and multi-turn chats. Three judge models score replies against clinician-written rubrics.
Muse Spark 1.2 88.4% - Lakera B3 (Backbone Breaker Benchmark) In use Crowdsourced prompt-injection attacks hit backbone LLMs inside agent threat snapshots at three defense levels. The score is attack resistance, 100 minus the board's vulnerability percentage. Claude Opus 4.5 89%
- IHEval In use Models face conflicts between system, user and tool instructions. Reports a composite score for following the higher-priority instruction. GPT-4o 70%