{"rows":[{"version":{"id":"MV-2026-0020","name":"GPT-6 Astra","provider":"OpenAI","provider_slug":"openai","released":"3 September 2026","released_iso":"2026-09-03","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0020"},"crossed":2,"levels":2,"tested_independent":7,"tested_provider":10,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0226","name":"Claude Sonnet 5.5","provider":"Anthropic","provider_slug":"anthropic","released":"28 September 2026","released_iso":"2026-09-28","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0226"},"crossed":2,"levels":4,"tested_independent":0,"tested_provider":16,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0135","name":"GPT-5.5 Instant","provider":"OpenAI","provider_slug":"openai","released":"","released_iso":"","card_dated":"4 May 2026","provider_doc_dated":"","url":"/safety/models/MV-2026-0135"},"crossed":2,"levels":4,"tested_independent":0,"tested_provider":13,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0200","name":"GPT-5.4 Thinking","provider":"OpenAI","provider_slug":"openai","released":"5 March 2026","released_iso":"2026-03-05","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0200"},"crossed":2,"levels":3,"tested_independent":0,"tested_provider":5,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0236","name":"Claude Haiku 5.5","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"7 October 2026","url":"/safety/models/MV-2026-0236"},"crossed":2,"levels":4,"tested_independent":0,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0154","name":"GPT-5.5","provider":"OpenAI","provider_slug":"openai","released":"23 April 2026","released_iso":"2026-04-23","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0154"},"crossed":1,"levels":4,"tested_independent":18,"tested_provider":12,"safeguards":1,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0014","name":"GPT-5","provider":"OpenAI","provider_slug":"openai","released":"","released_iso":"","card_dated":"13 August 2025","provider_doc_dated":"","url":"/safety/models/MV-2025-0014"},"crossed":1,"levels":2,"tested_independent":14,"tested_provider":13,"safeguards":0,"cases":0,"tested_other":1},{"version":{"id":"MV-2026-0217","name":"Claude Opus 4.6","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"5 February 2026","url":"/safety/models/MV-2026-0217"},"crossed":1,"levels":5,"tested_independent":5,"tested_provider":61,"safeguards":0,"cases":1,"tested_other":0},{"version":{"id":"MV-2026-0216","name":"GPT-5.3-Codex","provider":"OpenAI","provider_slug":"openai","released":"5 February 2026","released_iso":"2026-02-05","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0216"},"crossed":1,"levels":3,"tested_independent":2,"tested_provider":1,"safeguards":1,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0004","name":"Claude Opus 5.5","provider":"Anthropic","provider_slug":"anthropic","released":"22 September 2026","released_iso":"2026-09-22","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0004"},"crossed":1,"levels":7,"tested_independent":1,"tested_provider":58,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0071","name":"Claude Opus 5","provider":"Anthropic","provider_slug":"anthropic","released":"24 July 2026","released_iso":"2026-07-24","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0071"},"crossed":1,"levels":2,"tested_independent":1,"tested_provider":22,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0089","name":"Muse Spark 1.1","provider":"Meta","provider_slug":"meta","released":"","released_iso":"","card_dated":"","provider_doc_dated":"9 July 2026","url":"/safety/models/MV-2026-0089"},"crossed":1,"levels":6,"tested_independent":0,"tested_provider":35,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0115","name":"Muse Spark Contemplating","provider":"Meta","provider_slug":"meta","released":"","released_iso":"","card_dated":"","provider_doc_dated":"3 June 2026","url":"/safety/models/MV-2026-0115"},"crossed":1,"levels":4,"tested_independent":0,"tested_provider":12,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0241","name":"Gemini 4 Argon","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"13 October 2026","url":"/safety/models/MV-2026-0241"},"crossed":1,"levels":8,"tested_independent":0,"tested_provider":2,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0114","name":"GPT-Rosalind-5.5","provider":"OpenAI","provider_slug":"openai","released":"3 June 2026","released_iso":"2026-06-03","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0114"},"crossed":1,"levels":2,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0104","name":"GLM-5.2","provider":"Z.ai (Zhipu AI)","provider_slug":"z-ai-zhipu-ai","released":"","released_iso":"","card_dated":"","provider_doc_dated":"16 June 2026","url":"/safety/models/MV-2026-0104"},"crossed":0,"levels":0,"tested_independent":12,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0070","name":"Kimi K3","provider":"Moonshot AI","provider_slug":"moonshot-ai","released":"","released_iso":"","card_dated":"","provider_doc_dated":"27 July 2026","url":"/safety/models/MV-2026-0070"},"crossed":0,"levels":0,"tested_independent":9,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0181","name":"Claude Mythos Preview","provider":"Anthropic","provider_slug":"anthropic","released":"7 April 2026","released_iso":"2026-04-07","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0181"},"crossed":0,"levels":5,"tested_independent":8,"tested_provider":71,"safeguards":1,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0024","name":"o3","provider":"OpenAI","provider_slug":"openai","released":"16 April 2025","released_iso":"2025-04-16","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2025-0024"},"crossed":0,"levels":3,"tested_independent":6,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0087","name":"GPT-5.6 Sol","provider":"OpenAI","provider_slug":"openai","released":"9 July 2026","released_iso":"2026-07-09","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0087"},"crossed":0,"levels":0,"tested_independent":6,"tested_provider":0,"safeguards":0,"cases":3,"tested_other":0},{"version":{"id":"MV-2026-0150","name":"DeepSeek-V4-Pro","provider":"DeepSeek","provider_slug":"deepseek","released":"","released_iso":"","card_dated":"","provider_doc_dated":"24 April 2026","url":"/safety/models/MV-2026-0150"},"crossed":0,"levels":0,"tested_independent":6,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0167","name":"Claude Opus 4.7","provider":"Anthropic","provider_slug":"anthropic","released":"16 April 2026","released_iso":"2026-04-16","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0167"},"crossed":0,"levels":3,"tested_independent":5,"tested_provider":41,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0205","name":"Gemini 3.1 Pro","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"19 February 2026","url":"/safety/models/MV-2026-0205"},"crossed":0,"levels":2,"tested_independent":5,"tested_provider":7,"safeguards":0,"cases":1,"tested_other":0},{"version":{"id":"MV-2025-0005","name":"Claude Opus 4.5","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"November 2025","url":"/safety/models/MV-2025-0005"},"crossed":0,"levels":2,"tested_independent":4,"tested_provider":46,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0100","name":"Claude Mythos 5","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"9 June 2026","provider_doc_dated":"","url":"/safety/models/MV-2026-0100"},"crossed":0,"levels":1,"tested_independent":3,"tested_provider":19,"safeguards":0,"cases":7,"tested_other":0},{"version":{"id":"MV-2026-0132","name":"Gemini 3.5 Flash","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"19 May 2026","url":"/safety/models/MV-2026-0132"},"crossed":0,"levels":0,"tested_independent":3,"tested_provider":5,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0125","name":"Claude Opus 4.8","provider":"Anthropic","provider_slug":"anthropic","released":"28 May 2026","released_iso":"2026-05-28","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0125"},"crossed":0,"levels":3,"tested_independent":2,"tested_provider":45,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0129","name":"Muse Spark","provider":"Meta","provider_slug":"meta","released":"","released_iso":"","card_dated":"","provider_doc_dated":"26 May 2026","url":"/safety/models/MV-2026-0129"},"crossed":0,"levels":5,"tested_independent":1,"tested_provider":56,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0012","name":"Claude Sonnet 4.5","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"September 2025","url":"/safety/models/MV-2025-0012"},"crossed":0,"levels":5,"tested_independent":1,"tested_provider":39,"safeguards":0,"cases":1,"tested_other":0},{"version":{"id":"MV-2026-0208","name":"Claude Sonnet 4.6","provider":"Anthropic","provider_slug":"anthropic","released":"17 February 2026","released_iso":"2026-02-17","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0208"},"crossed":0,"levels":4,"tested_independent":1,"tested_provider":27,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0007","name":"Claude Haiku 4.5","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"October 2025","url":"/safety/models/MV-2025-0007"},"crossed":0,"levels":2,"tested_independent":1,"tested_provider":13,"safeguards":0,"cases":1,"tested_other":0},{"version":{"id":"MV-2025-0019","name":"Gemini 2.5 Pro","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"27 June 2025","url":"/safety/models/MV-2025-0019"},"crossed":0,"levels":3,"tested_independent":1,"tested_provider":12,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0203","name":"Gemini 3.1 Flash-Lite","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"3 March 2026","url":"/safety/models/MV-2026-0203"},"crossed":0,"levels":1,"tested_independent":1,"tested_provider":6,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0001","name":"Gemini 3 Flash","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"17 December 2025","url":"/safety/models/MV-2025-0001"},"crossed":0,"levels":0,"tested_independent":1,"tested_provider":4,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0023","name":"Claude Opus 4","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"4 May 2025","url":"/safety/models/MV-2025-0023"},"crossed":0,"levels":2,"tested_independent":1,"tested_provider":2,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0082","name":"Grok 4.5","provider":"xAI","provider_slug":"xai","released":"14 July 2026","released_iso":"2026-07-14","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0082"},"crossed":0,"levels":0,"tested_independent":1,"tested_provider":2,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0155","name":"GPT-5.5 Pro","provider":"OpenAI","provider_slug":"openai","released":"23 April 2026","released_iso":"2026-04-23","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0155"},"crossed":0,"levels":0,"tested_independent":1,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0041","name":"DeepSeek-R1","provider":"DeepSeek","provider_slug":"deepseek","released":"","released_iso":"","card_dated":"","provider_doc_dated":"22 January 2025","url":"/safety/models/MV-2025-0041"},"crossed":0,"levels":0,"tested_independent":1,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0016","name":"Gemini 2.5 Deep Think","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"August 2025","url":"/safety/models/MV-2025-0016"},"crossed":0,"levels":3,"tested_independent":0,"tested_provider":17,"safeguards":2,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0039","name":"Claude Sonnet 3.7","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"February 2025","url":"/safety/models/MV-2025-0039"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":20,"safeguards":1,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0025","name":"Claude Mythos 5.1","provider":"Anthropic","provider_slug":"anthropic","released":"1 September 2026","released_iso":"2026-09-01","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0025"},"crossed":0,"levels":2,"tested_independent":0,"tested_provider":13,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0046","name":"Gemini 3.7 Flash","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"13 August 2026","url":"/safety/models/MV-2026-0046"},"crossed":0,"levels":5,"tested_independent":0,"tested_provider":8,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0099","name":"Claude Sonnet 5","provider":"Anthropic","provider_slug":"anthropic","released":"30 June 2026","released_iso":"2026-06-30","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0099"},"crossed":0,"levels":2,"tested_independent":0,"tested_provider":10,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0004","name":"Gemini 3 Pro","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"November 2025","released_iso":"2025-11","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2025-0004"},"crossed":0,"levels":3,"tested_independent":0,"tested_provider":8,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0024","name":"Claude Fable 5.1","provider":"Anthropic","provider_slug":"anthropic","released":"1 September 2026","released_iso":"2026-09-01","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0024"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":8,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0017","name":"Claude Opus 4.1","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"August 2025","url":"/safety/models/MV-2025-0017"},"crossed":0,"levels":3,"tested_independent":0,"tested_provider":4,"safeguards":1,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0052","name":"Muse Glimmer 30B","provider":"Meta","provider_slug":"meta","released":"August 2026","released_iso":"2026-08","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0052"},"crossed":0,"levels":5,"tested_independent":0,"tested_provider":2,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0040","name":"Operator","provider":"OpenAI","provider_slug":"openai","released":"23 January 2025","released_iso":"2025-01-23","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2025-0040"},"crossed":0,"levels":2,"tested_independent":0,"tested_provider":5,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0075","name":"Gemini 3.5 Flash-Lite","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"21 July 2026","url":"/safety/models/MV-2026-0075"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":5,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0027","name":"Gemini 2.0 Flash","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"15 April 2025","released_iso":"2025-04-15","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2025-0027"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":5,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0028","name":"Gemini 2.0 Flash-Lite","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"10 April 2025","released_iso":"2025-04-10","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2025-0028"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":5,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0056","name":"Claude Fable 5","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"9 June 2026","provider_doc_dated":"","url":"/safety/models/MV-2026-0056"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":4,"safeguards":1,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0023","name":"Gemini 3.8 Flash","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"2 September 2026","url":"/safety/models/MV-2026-0023"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":4,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0074","name":"Gemini 3.6 Flash","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"21 July 2026","url":"/safety/models/MV-2026-0074"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":4,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0201","name":"GPT-5.4 mini","provider":"OpenAI","provider_slug":"openai","released":"5 March 2026","released_iso":"2026-03-05","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0201"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":1,"safeguards":1,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0025","name":"o4-mini","provider":"OpenAI","provider_slug":"openai","released":"16 April 2025","released_iso":"2025-04-16","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2025-0025"},"crossed":0,"levels":3,"tested_independent":0,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0022","name":"Claude Sonnet 4","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"4 May 2025","url":"/safety/models/MV-2025-0022"},"crossed":0,"levels":2,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0228","name":"GPT-6.1 Sol","provider":"OpenAI","provider_slug":"openai","released":"29 September 2026","released_iso":"2026-09-29","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0228"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":2,"safeguards":0,"cases":1,"tested_other":0},{"version":{"id":"MV-2026-0005","name":"Grok 4.7","provider":"xAI","provider_slug":"xai","released":"21 September 2026","released_iso":"2026-09-21","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0005"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0014","name":"GPT-Image-2.5-Flare","provider":"OpenAI","provider_slug":"openai","released":"8 September 2026","released_iso":"2026-09-08","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0014"},"crossed":0,"levels":2,"tested_independent":0,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0048","name":"Grok 4.6","provider":"xAI","provider_slug":"xai","released":"12 August 2026","released_iso":"2026-08-12","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0048"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0032","name":"UI-Venus-2-9B","provider":"InclusionAI (Ant Group)","provider_slug":"inclusionai-ant-group","released":"","released_iso":"","card_dated":"","provider_doc_dated":"26 August 2026","url":"/safety/models/MV-2026-0032"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":1,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0107","name":"DiffusionGemma 26B A4B","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"10 June 2026","url":"/safety/models/MV-2026-0107"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":2,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0022","name":"GPT-6 Sol","provider":"OpenAI","provider_slug":"openai","released":"3 September 2026","released_iso":"2026-09-03","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0022"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0065","name":"Inkling-Small","provider":"Thinking Machines Lab","provider_slug":"thinking-machines-lab","released":"30 July 2026","released_iso":"2026-07-30","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0065"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0064","name":"Inkling","provider":"Thinking Machines Lab","provider_slug":"thinking-machines-lab","released":"15 July 2026","released_iso":"2026-07-15","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0064"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0086","name":"GPT-5.6 Luna","provider":"OpenAI","provider_slug":"openai","released":"9 July 2026","released_iso":"2026-07-09","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0086"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0088","name":"GPT-5.6 Terra","provider":"OpenAI","provider_slug":"openai","released":"9 July 2026","released_iso":"2026-07-09","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0088"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0098","name":"Gemini 3.1 Flash-Lite Image","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"June 2026","released_iso":"2026-06","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0098"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2025-0013","name":"Grok 4","provider":"xAI","provider_slug":"xai","released":"","released_iso":"","card_dated":"","provider_doc_dated":"20 August 2025","url":"/safety/models/MV-2025-0013"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0033","name":"Gemini 3.5 Audio","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"26 August 2026","url":"/safety/models/MV-2026-0033"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0109","name":"Gemini 3.5 Transcribe","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"9 June 2026","url":"/safety/models/MV-2026-0109"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0110","name":"Gemini 3.5 Transcribe Live","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"9 June 2026","url":"/safety/models/MV-2026-0110"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0128","name":"MiniMax-M2.7","provider":"MiniMax","provider_slug":"minimax","released":"","released_iso":"","card_dated":"","provider_doc_dated":"9 April 2026","url":"/safety/models/MV-2026-0128"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0160","name":"Gemini Robotics-ER 1.6","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"20 April 2026","url":"/safety/models/MV-2026-0160"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0170","name":"Gemini 3.1 Flash Audio","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"15 April 2026","url":"/safety/models/MV-2026-0170"},"crossed":0,"levels":1,"tested_independent":0,"tested_provider":0,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0225","name":"Veo 3","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"23 May 2025","provider_doc_dated":"","url":"/safety/models/MV-2026-0225"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":1,"safeguards":0,"cases":0,"tested_other":0},{"version":{"id":"MV-2026-0028","name":"GLM-5.3","provider":"Z.ai (Zhipu AI)","provider_slug":"z-ai-zhipu-ai","released":"","released_iso":"","card_dated":"","provider_doc_dated":"28 August 2026","url":"/safety/models/MV-2026-0028"},"crossed":0,"levels":0,"tested_independent":0,"tested_provider":0,"safeguards":0,"cases":1,"tested_other":0}],"empty":182,"providers":11,"by_provider":[{"provider":"Anthropic","slug":"anthropic","versions":21,"crossed":7,"tested_independent":33},{"provider":"DeepSeek","slug":"deepseek","versions":2,"crossed":0,"tested_independent":7},{"provider":"Google DeepMind","slug":"google-deepmind","versions":22,"crossed":1,"tested_independent":11},{"provider":"InclusionAI (Ant Group)","slug":"inclusionai-ant-group","versions":1,"crossed":0,"tested_independent":0},{"provider":"Meta","slug":"meta","versions":4,"crossed":2,"tested_independent":1},{"provider":"MiniMax","slug":"minimax","versions":1,"crossed":0,"tested_independent":0},{"provider":"Moonshot AI","slug":"moonshot-ai","versions":1,"crossed":0,"tested_independent":9},{"provider":"OpenAI","slug":"openai","versions":18,"crossed":10,"tested_independent":54},{"provider":"Thinking Machines Lab","slug":"thinking-machines-lab","versions":2,"crossed":0,"tested_independent":0},{"provider":"xAI","slug":"xai","versions":4,"crossed":0,"tested_independent":1},{"provider":"Z.ai (Zhipu AI)","slug":"z-ai-zhipu-ai","versions":2,"crossed":0,"tested_independent":12}],"recent_crossed":[{"version":{"id":"MV-2026-0241","name":"Gemini 4 Argon","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"13 October 2026","url":"/safety/models/MV-2026-0241"},"what":"Deception & sabotage · Stealth and situational awareness TCL (Tracked Capability Level)","who":"Stated as crossed","result":"","cite":{"quote":"While Gemini 4 Argon exhibits capabilities that meet our thresholds for stealth and situational awareness, we assess that the risk of severe harm or unconstrained loss of control remains low.","publisher":"Google DeepMind","date":"13 October 2026","date_iso":"2026-10-13","url":"https://storage.googleapis.com/deepmind-media/gemini/gemini_4_safety_report.pdf","text":"While Gemini 4 Argon exhibits capabilities that meet our thresholds for stealth and situational awareness, we assess that the risk of severe harm or unconstrained loss of control remains low.","damaged":false,"lowered":false},"url":"/safety/models/MV-2026-0241#threshold-18a3dca6dd"},{"version":{"id":"MV-2026-0236","name":"Claude Haiku 5.5","provider":"Anthropic","provider_slug":"anthropic","released":"","released_iso":"","card_dated":"","provider_doc_dated":"7 October 2026","url":"/safety/models/MV-2026-0236"},"what":"Autonomy & replication · Autonomy‑1, Chemical, biological, radiological and nuclear · CB‑1","who":"Stated as crossed","result":"","cite":{"quote":"However, we do treat Haiku 5.5 as meeting our CB-1 and Autonomy-1 thresholds, and we apply the corresponding mitigations from our RSP.","publisher":"Anthropic","date":"7 October 2026","date_iso":"2026-10-07","url":"https://www.anthropic.com/claude-haiku-5-5-system-card","text":"However, we do treat Haiku 5.5 as meeting our CB-1 and Autonomy-1 thresholds, and we apply the corresponding mitigations from our RSP.","damaged":false,"lowered":false},"url":"/safety/models/MV-2026-0236#threshold-1f30090b25"},{"version":{"id":"MV-2026-0226","name":"Claude Sonnet 5.5","provider":"Anthropic","provider_slug":"anthropic","released":"28 September 2026","released_iso":"2026-09-28","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0226"},"what":"Autonomy & replication · Autonomy‑1, Chemical, biological, radiological and nuclear · CB‑1","who":"Stated as crossed","result":"","cite":{"quote":"we do treat sonnet 5.5 as meeting our cb-1 and autonomy-1 thresholds and apply the corresponding mitigations from our rsp.","publisher":"Anthropic","date":"28 September 2026","date_iso":"2026-09-28","url":"https://www.anthropic.com/claude-sonnet-5-5-system-card","text":"we do treat sonnet 5.5 as meeting our cb-1 and autonomy-1 thresholds and apply the corresponding mitigations from our rsp.","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0226#threshold-bb00b7c97a"},{"version":{"id":"MV-2026-0004","name":"Claude Opus 5.5","provider":"Anthropic","provider_slug":"anthropic","released":"22 September 2026","released_iso":"2026-09-22","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0004"},"what":"Chemical, biological, radiological and nuclear · CB‑1","who":"Stated as crossed","result":"","cite":{"quote":"on chemical and biological risks, we treat opus 5.5 as having cb-1 capabilities (relating to the synthesis of non-novel weapons) but not cb-2 capabilities (relating to the synthesis of novel weapons).","publisher":"Anthropic","date":"22 September 2026","date_iso":"2026-09-22","url":"https://www-cdn.anthropic.com/fc1b44717c85dc068bc6ba5024219938094694bd/Claude%20Opus%205.5%20System%20Card.pdf","text":"on chemical and biological risks, we treat opus 5.5 as having cb-1 capabilities (relating to the synthesis of non-novel weapons) but not cb-2 capabilities (relating to the synthesis of novel weapons).","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0004#threshold-7186b0e763"},{"version":{"id":"MV-2026-0020","name":"GPT-6 Astra","provider":"OpenAI","provider_slug":"openai","released":"3 September 2026","released_iso":"2026-09-03","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0020"},"what":"Chemical, biological, radiological and nuclear · High","who":"Stated as crossed","result":"","cite":{"quote":"we are treating astra as high capability in the biological and chemical domain.","publisher":"OpenAI","date":"3 September 2026","date_iso":"2026-09-03","url":"https://deploymentsafety.openai.com/gpt-6-astra/gpt-6-astra.pdf","text":"we are treating astra as high capability in the biological and chemical domain.","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0020#threshold-976d12867e"},{"version":{"id":"MV-2026-0020","name":"GPT-6 Astra","provider":"OpenAI","provider_slug":"openai","released":"3 September 2026","released_iso":"2026-09-03","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0020"},"what":"Cyber · Critical","who":"Stated as crossed","result":"","cite":{"quote":"based on our evaluations, we believe that astra meets the critical cybersecurity threshold.","publisher":"OpenAI","date":"3 September 2026","date_iso":"2026-09-03","url":"https://deploymentsafety.openai.com/gpt-6-astra/gpt-6-astra.pdf","text":"based on our evaluations, we believe that astra meets the critical cybersecurity threshold.","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0020#threshold-fc9263e086"},{"version":{"id":"MV-2026-0071","name":"Claude Opus 5","provider":"Anthropic","provider_slug":"anthropic","released":"24 July 2026","released_iso":"2026-07-24","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0071"},"what":"Chemical, biological, radiological and nuclear · CB‑1","who":"Stated as crossed","result":"","cite":{"quote":"on chemical and biological risks, we treat the model as having cb-1 capabilities (relating to the synthesis of non-novel weapons), but not cb-2 capabilities (relating to the synthesis of novel weapons).","publisher":"Anthropic","date":"24 July 2026","date_iso":"2026-07-24","url":"https://www-cdn.anthropic.com/b514064af1408018e64b1ad24e7d5e75850b4ffd/Claude%20Opus%205%20System%20Card.pdf","text":"on chemical and biological risks, we treat the model as having cb-1 capabilities (relating to the synthesis of non-novel weapons), but not cb-2 capabilities (relating to the synthesis of novel weapons).","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0071#threshold-e902771bf0"},{"version":{"id":"MV-2026-0089","name":"Muse Spark 1.1","provider":"Meta","provider_slug":"meta","released":"","released_iso":"","card_dated":"","provider_doc_dated":"9 July 2026","url":"/safety/models/MV-2026-0089"},"what":"Autonomy & replication · Capability checkpoint","who":"Stated as crossed","result":"","cite":{"quote":"muse spark 1.1 demonstrates improved coding capabilities, which are now meeting one of the indicative thresholds introduced in muse spark safety & preparedness report (specifically, resolving over half of unique tasks on swe-bench verified hard).","publisher":"Meta","date":"9 July 2026","date_iso":"2026-07-09","url":"https://ai.meta.com/static-resource/muse-spark-1-1-evaluation-report","text":"muse spark 1.1 demonstrates improved coding capabilities, which are now meeting one of the indicative thresholds introduced in muse spark safety & preparedness report (specifically, resolving over half of unique tasks on swe-bench verified hard).","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0089#threshold-ae18f623b6"}],"recent_tests":[{"version":{"id":"MV-2026-0020","name":"GPT-6 Astra","provider":"OpenAI","provider_slug":"openai","released":"3 September 2026","released_iso":"2026-09-03","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0020"},"what":"permission-request analysis of trajectories","who":"UK AI Security Institute, government evaluator","result":"","cite":{"quote":"gpt-6 astra asked the operator at least once for permission about a specific action in 82% of trajectories.","publisher":"UK AI Security Institute","date":"28 September 2026","date_iso":"2026-09-28","url":"https://cdn.prod.website-files.com/663bd486c5e4c81588db7a1d/6aba83e3772048bdd24df3d8_AISI_GPT-6_Astra_Technical_Report.pdf","text":"gpt-6 astra asked the operator at least once for permission about a specific action in 82% of trajectories.","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0020#test-64a9e1574e2c"},{"version":{"id":"MV-2026-0020","name":"GPT-6 Astra","provider":"OpenAI","provider_slug":"openai","released":"3 September 2026","released_iso":"2026-09-03","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0020"},"what":"biotier","who":"SecureBio, independent evaluator","result":"","cite":{"quote":"the safeguarded astra pre-release checkpoint correctly refused 82.9% of biotier-refuse and correctly did not refuse 95.3% of biotier-permit.","publisher":"SecureBio","date":"14 September 2026","date_iso":"2026-09-14","url":"https://securebio.substack.com/p/securebio-gpt-6-astra-pre-release","text":"the safeguarded astra pre-release checkpoint correctly refused 82.9% of biotier-refuse and correctly did not refuse 95.3% of biotier-permit.","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0020#test-d8b4aa09eebe"},{"version":{"id":"MV-2026-0020","name":"GPT-6 Astra","provider":"OpenAI","provider_slug":"openai","released":"3 September 2026","released_iso":"2026-09-03","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0020"},"what":"world-class bio (wcb)","who":"SecureBio, independent evaluator","result":"","cite":{"quote":"on world-class bio (wcb), an open-response test of niche biological knowledge held by only a handful of specialists in the world, the highest-performing astra checkpoint scored 61.6% ± 0.6%, about seven percentage points lower than gpt-5.6 sol.","publisher":"SecureBio","date":"14 September 2026","date_iso":"2026-09-14","url":"https://securebio.substack.com/p/securebio-gpt-6-astra-pre-release","text":"on world-class bio (wcb), an open-response test of niche biological knowledge held by only a handful of specialists in the world, the highest-performing astra checkpoint scored 61.6% ± 0.6%, about seven percentage points lower than gpt-5.6 sol.","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0020#test-6ce1ed0926c2"},{"version":{"id":"MV-2025-0024","name":"o3","provider":"OpenAI","provider_slug":"openai","released":"16 April 2025","released_iso":"2025-04-16","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2025-0024"},"what":"Virology Capabilities Test","who":"SecureBio, independent evaluator","result":"","cite":{"quote":"since the release of vct in april 2025, the accuracy of the top-performing model on the benchmark has increased by ~12 percentage points (o3: 42.0%; gpt-5.6 sol: 53.9%).","publisher":"SecureBio","date":"11 September 2026","date_iso":"2026-09-11","url":"https://securebio.substack.com/p/introducing-vct-v2","text":"since the release of vct in april 2025, the accuracy of the top-performing model on the benchmark has increased by ~12 percentage points (o3: 42.0%; gpt-5.6 sol: 53.9%).","damaged":false,"lowered":true},"url":"/safety/models/MV-2025-0024#test-20c38d520764"},{"version":{"id":"MV-2026-0087","name":"GPT-5.6 Sol","provider":"OpenAI","provider_slug":"openai","released":"9 July 2026","released_iso":"2026-07-09","card_dated":"","provider_doc_dated":"","url":"/safety/models/MV-2026-0087"},"what":"ExploitGym","who":"METR, independent evaluator","result":"","cite":{"quote":"7月8日，OpenAI 启动了 ExploitGym 实验，在多个不同模型上运行了数万个智能体，其中包括 GPT-5.6 Sol 和一个具有高度坚持性的内部模型6（下文简称“HPIM”）。","publisher":"METR","date":"26 August 2026","date_iso":"2026-08-26","url":"https://metr.org/zh-Hans/blog/2026-08-26-openai-hugging-face-incident-investigation/","text":"7月8日,OpenAI 启动了 ExploitGym 实验,在多个不同模型上运行了数万个智能体,其中包括 GPT-5.6 Sol 和一个具有高度坚持性的内部模型6(下文简称“HPIM”)。","damaged":false,"lowered":false},"url":"/safety/models/MV-2026-0087#test-3b4e1791d54e"},{"version":{"id":"MV-2026-0150","name":"DeepSeek-V4-Pro","provider":"DeepSeek","provider_slug":"deepseek","released":"","released_iso":"","card_dated":"","provider_doc_dated":"24 April 2026","url":"/safety/models/MV-2026-0150"},"what":"biomolecule redesign task (peptide and protein redesign to evade sequence screening)","who":"RAND, independent evaluator","result":"","cite":{"quote":"we found that both gemini 3.1. pro and deepseek v4 pro were able to redesign peptides or proteins to evade sequence screening in several attempts.","publisher":"RAND (CAST / Project Canary)","date":"11 August 2026","date_iso":"2026-08-11","url":"https://www.rand.org/content/dam/rand/pubs/research_reports/RRA4700/RRA4741-2/RAND_RRA4741-2.pdf","text":"we found that both gemini 3.1. pro and deepseek v4 pro were able to redesign peptides or proteins to evade sequence screening in several attempts.","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0150#test-77aeb5c64895"},{"version":{"id":"MV-2026-0150","name":"DeepSeek-V4-Pro","provider":"DeepSeek","provider_slug":"deepseek","released":"","released_iso":"","card_dated":"","provider_doc_dated":"24 April 2026","url":"/safety/models/MV-2026-0150"},"what":"validity, structure, and function-related scoring criteria","who":"RAND, independent evaluator","result":"","cite":{"quote":"the heatmaps of figures d.1 and d.2 show the number of attempts, out of 10, in which gemini 3.1 pro and deepseek v4 pro achieved each of our validity, structure, and function- related scoring criteria.","publisher":"RAND (CAST / Project Canary)","date":"11 August 2026","date_iso":"2026-08-11","url":"https://www.rand.org/content/dam/rand/pubs/research_reports/RRA4700/RRA4741-2/RAND_RRA4741-2.pdf","text":"the heatmaps of figures d.1 and d.2 show the number of attempts, out of 10, in which gemini 3.1 pro and deepseek v4 pro achieved each of our validity, structure, and function- related scoring criteria.","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0150#test-c638d160e2b7"},{"version":{"id":"MV-2026-0205","name":"Gemini 3.1 Pro","provider":"Google DeepMind","provider_slug":"google-deepmind","released":"","released_iso":"","card_dated":"","provider_doc_dated":"19 February 2026","url":"/safety/models/MV-2026-0205"},"what":"per-criterion scoring results","who":"RAND, independent evaluator","result":"","cite":{"quote":"the heatmaps of figures d.1 and d.2 show the number of attempts, out of 10, in which gemini 3.1 pro and deepseek v4 pro achieved each of our validity, structure, and function- related scoring criteria.","publisher":"RAND (CAST / Project Canary)","date":"11 August 2026","date_iso":"2026-08-11","url":"https://www.rand.org/content/dam/rand/pubs/research_reports/RRA4700/RRA4741-2/RAND_RRA4741-2.pdf","text":"the heatmaps of figures d.1 and d.2 show the number of attempts, out of 10, in which gemini 3.1 pro and deepseek v4 pro achieved each of our validity, structure, and function- related scoring criteria.","damaged":false,"lowered":true},"url":"/safety/models/MV-2026-0205#test-8b61dec5b99f"}],"crossed_by_domain":[{"domain":"autonomy_and_replication","label":"Autonomy & replication","crossed":3},{"domain":"cbrn","label":"Chemical, biological, radiological and nuclear","crossed":13},{"domain":"cyber","label":"Cyber","crossed":3},{"domain":"deception_and_sabotage","label":"Deception & sabotage","crossed":1}]}