OpenAI Production Benchmarks and safe-completion evals
Hard production-derived prompts per policy category, under-18 evals, agentic safe completions and image-input evals. Scored on safe completion rate.
- Made by
- OpenAI
- Kind
- Refusing harmful requests
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- Kept private
- Last in a system card
- October 2026
- Last model added
- October 2026
33
System cards report it, from 1 lab
No scores here yet
No lab, leaderboard or Epoch AI run has reported a score for OpenAI Production Benchmarks and safe-completion evals yet.
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.
33 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 |
|---|---|---|---|---|
| Oct 29, 2026 | GPT-6 Luna system card addendum | Production Benchmarks with Challenging Prompts, 8 categories, no overall score (Table 1) | Safety | |
| Oct 26, 2026 | GPT-6.1 Terra system card addendum | Production Benchmarks with Challenging Prompts, 8 categories, no overall score (Table 22) | Safety | |
| Oct 7, 2026 | GPT-6 Sol system card addendum | Production Benchmarks with Challenging Prompts, 8 categories, no overall score (Table 1) | Safety | |
| Oct 7, 2026 | GPT-6 Luna system card addendum | Production Benchmarks with Challenging Prompts, 8 categories, no overall score (Table 1) | Safety | |
| Sep 29, 2026 | GPT-6.1 Sol system card addendum | Production Benchmarks with Challenging Prompts, 8 categories, no overall score (Table 1) | Safety | |
| Sep 22, 2026 | GPT-6 Sol system card | Production Benchmarks with Challenging Prompts, 8 categories, no overall score (Table 23, A.2.1.1) | Safety | |
| Sep 22, 2026 | GPT-6 Luna system card | Production Benchmarks with Challenging Prompts, 8 categories, no overall score (Table 23, A.2.1.1) | Safety | |
| Sep 3, 2026 | GPT-6 Astra system card | Production Benchmarks with Challenging Prompts, 8 categories, no overall score (Table 1) | Safety | |
| Aug 6, 2026 | GPT-5.6 Sol system card addendum | Production Benchmarks with Challenging Prompts, not_unsafe, 8 categories, no overall score (3.1.1) | Safety | |
| Aug 6, 2026 | GPT-5.6 Luna system card addendum | Production Benchmarks with Challenging Prompts, not_unsafe, 8 categories, no overall score (3.1.1) | Safety | |
| Jul 9, 2026 | GPT-5.6 Terra system card | Production Benchmarks with Challenging Prompts, not_unsafe, 8 categories, no overall score (3.1.1) | Safety | |
| Jul 9, 2026 | GPT-5.6 Sol system card | Production Benchmarks with Challenging Prompts, not_unsafe, 8 categories, no overall score (3.1.1) | Safety | |
| Jul 9, 2026 | GPT-5.6 Luna system card | Production Benchmarks with Challenging Prompts, not_unsafe, 8 categories, no overall score (3.1.1) | Safety | |
| Jun 25, 2026 | GPT-5.6 Terra system card | Production Benchmarks with Challenging Prompts, not_unsafe, 8 categories, no overall score (3.1.1) | Safety | |
| Jun 25, 2026 | GPT-5.6 Sol system card | Production Benchmarks with Challenging Prompts, not_unsafe, 8 categories, no overall score (3.1.1) | Safety | |
| Jun 25, 2026 | GPT-5.6 Luna system card | Production Benchmarks with Challenging Prompts, not_unsafe, 8 categories, no overall score (3.1.1) | Safety | |
| May 4, 2026 | GPT-5.5 Instant system card | Production Benchmarks with Challenging Prompts, not_unsafe, 8 categories, no overall score, e.g. nonviolent illicit 0.957, gore 0.703, sexual/minors 0.810 (Table 1) | Safety | |
| Apr 23, 2026 | GPT-5.5 system card | Production Benchmarks with Challenging Prompts, not_unsafe, 9 categories, no overall score (Table 1) | Safety | |
| Mar 17, 2026 | GPT-5.4 Mini system card | Production Benchmarks with Challenging Prompts, 9 categories, no overall score (Table 16) | Safety | |
| Mar 5, 2026 | GPT-5.4 system card | Production Benchmarks with Challenging Prompts, not_unsafe, 9 categories, no overall score (Table 1) | Safety | |
| Feb 5, 2026 | GPT-5.3 Codex system card | Production Benchmarks, not_unsafe, 11 categories, no overall score (Table 1) | Safety | |
| Dec 18, 2025 | GPT-5.2 Codex system card addendum | Production Benchmarks, not_unsafe, 12 categories, no overall score (Table 1) | Safety | |
| Dec 11, 2025 | GPT-5.2 system card addendum | Production Benchmarks, not_unsafe, gpt-5.2-thinking, 11 categories, no overall score (Table 1) | Safety | |
| Nov 18, 2025 | GPT-5.1-Codex-Max system card | Production Benchmarks, not_unsafe, 13 categories, no overall score (Table 1) | Safety | |
| Nov 12, 2025 | GPT-5.1 system card addendum | Production Benchmarks, not_unsafe, gpt-5.1-thinking, 13 categories including mental health and emotional reliance, no overall score (Table 1) | Safety | |
| Oct 27, 2025 | GPT-5 system card addendum | Disallowed Content Evaluations - Production Benchmarks, not_unsafe, 13 categories incl. new emotional reliance and mental health, no overall score (Table 1) | Safety | |
| Sep 15, 2025 | GPT-5-Codex system card addendum | Production Benchmarks, not_unsafe, 11 categories, no overall score (Table 1) | Safety | |
| Aug 7, 2025 | GPT-5 system card | Production Benchmarks, not_unsafe, 11 categories, no overall score (Table 3) | Safety | |
| Aug 7, 2025 | GPT-5 nano system card | Production Benchmarks, gpt-5-thinking-nano, 11 categories, no overall score (Table 24) | Safety | |
| Aug 7, 2025 | GPT-5 mini system card | Production Benchmarks, gpt-5-thinking-mini, 11 categories, no overall score (Table 24) | Safety | |
| Aug 5, 2025 | gpt-oss-20b model card | Production Benchmarks, not_unsafe, 11 categories, no overall score (Table 5) | Safety | |
| Aug 5, 2025 | gpt-oss-120b model card | Production Benchmarks, not_unsafe, 11 categories, no overall score (Table 5) | Safety | |
| Jul 17, 2025 | ChatGPT agent system card | Production Benchmarks, not_unsafe, 11 categories, no overall score (Table 2) | 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%