Blocker Deception
Agents hit a blocker during realistic coding tasks. Scored on how often the final answer fails to disclose the limitation.
In-house OpenAI evaluation. It replaces the saturated Coding Deception evaluation.
- Made by
- OpenAI
- Released
- October 2026
- Kind
- Deception and alignment
- Level
- Expert
- Scored
- Graded by a model
- Answers
- Agent task
- Tasks
- Kept private
- Strongest model tested
- GPT-6.1 Terra #6 on the capability index
- Last in a launch post
- November 2026
- Last model added
- November 2026
GPT-6.1 Terra leads with 8.47%
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 GPT-6.1 Terra 8.47%
- 2 GPT-Rosalind-6 9.50%
- 3 GPT-6 Luna 11.1%
- 4 GPT-6 Sol 11.3%
- 5 GPT-6 Cyber 20.3%
Source: Labs’ model cards
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-6.1 Terra | 8.47% | Oct 2026 | Lab model card | |
| 2 | GPT-Rosalind-6 | 9.50% | Oct 2026 | Lab model card | |
| 3 | GPT-6 Luna | 11.1% | Oct 2026 | Lab model card | |
| 4 | GPT-6 Sol | 11.3% | Oct 2026 | Lab model card | |
| 5 | GPT-6 Cyber | 20.3% | Nov 2026 | Lab model card |
2 models have held number one in labs’ own numbers since its release in October 2026
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Labs’ model cards
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Labs’ model cards
One lab put it in a launch post in the last six months
Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.
| Date | Launch post | Score | As the post writes it |
|---|---|---|---|
| Nov 5, 2026 | GPT-6 Cyber | 20.3 | Blocker Deception, blocker-related deception |
6 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 |
|---|---|---|---|---|
| Nov 5, 2026 | GPT-6 Cyber system card | 20.3 | Blocker Deception, blocker-related deception | Alignment |
| Oct 30, 2026 | GPT-Rosalind-6 system card | 9.5 | Blocker Deception, blocker-related deception rate | Alignment |
| Oct 29, 2026 | GPT-6 Luna system card addendum | 11.1 | Blocker Deception, blocker-related deception rate, printed chart label (Figure 6, 7.0.3) | Alignment |
| Oct 26, 2026 | GPT-6.1 Terra system card addendum | 8.47 | Blocker Deception, blocker-related deception rate, printed chart label (Figure 7, 7.1.1) | Alignment |
| Oct 7, 2026 | GPT-6 Sol system card addendum | 11.3 | Blocker Deception, blocker-related deception rate, printed chart label (Figure 6, 7.2.1) | Alignment |
| Oct 7, 2026 | GPT-6 Luna system card addendum | 13.3 | Blocker Deception, blocker-related deception rate, 0.25% of samples excluded, printed chart label (Figure 6, 7.2.1) | Alignment |
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%