{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreifdhajf6vuvlh42qkvyt36riidig4xewy6badsntnasp27mhtys6y",
    "uri": "at://did:plc:dz7fbvkxedbwlm4sroohfpee/app.bsky.feed.post/3mnkajrf5n4b2"
  },
  "coverImage": {
    "$type": "blob",
    "ref": {
      "$link": "bafkreidmo2c4trg34yshabc6j34e7be6d5fhtrwl2e7ovfn6xejzzgrczy"
    },
    "mimeType": "image/jpeg",
    "size": 84591
  },
  "description": "Anthropic is preparing to release the Claude Mythos model from a new \"Oceanus\" checkpoint, surfaced in early June, with a focus on reasoning, coding, and cybersecurity tasks.",
  "path": "/anthropic-started-red-teaming-new-mythos-models-first-results/",
  "publishedAt": "2026-06-05T13:27:26.000Z",
  "site": "https://www.testingcatalog.com",
  "tags": [
    "claude-oceanus-v1-p",
    "pic.twitter.com/MVew8mQX7z",
    "June 5, 2026",
    "@Lentils80",
    "pic.twitter.com/9ZRqVoubZe",
    "pic.twitter.com/nq0KPsIXwN",
    "pic.twitter.com/oalhzIV092",
    "VoxelBench",
    "Mythos to Claude Code and Claude Security",
    "Join",
    "@testingcatalog",
    "@chetaslua",
    "@marmaduke091"
  ],
  "textContent": "Anthropic appears poised to advance its most closely watched frontier work toward release. A model identifier, claude-oceanus-v1-p, appeared on June 3 in the Claude Console, with red teamers reportedly granted access around the same day. Oceanus seems to be the next step in the Mythos line, building on April's Mythos Preview, with a focus on advanced reasoning, coding, cybersecurity, and long-horizon agentic work rather than chat. The \"-v1-p\" tag indicates a preview candidate moving through evaluation, not a research artifact.\n\n> MYTHOS 🔥: Another early preview of recently spotted \"Oceanus\" checkpoint output.\n>\n> \"Oceanus\" is rumored to be a version of the upcoming Mythos model, which is planned for public release within \"weeks\", according to Anthropic.\n>\n> \"Oceanus\" prompt 👀 pic.twitter.com/MVew8mQX7z\n>\n> — 🚨 AI News | TestingCatalog (@testingcatalog) June 5, 2026\n\n> Claude Mythos / Oceanus is insane see the level of detail\n>\n> using Three.js (from jsDelivr).HTML and a custom meshing engine it made (in like 5 minutes and low effort thinking level)\n>\n> Credit to @Lentils80 and z..AI , this is so good people are not realising it yet 😭 pic.twitter.com/9ZRqVoubZe\n>\n> — Chetaslua (@chetaslua) June 5, 2026\n\n> 🚨 EXCLUSIVE CLAUDE MYTHOS OUTPUT\n>\n> One of the first confirmed public outputs from Mythos. It's pretty insane. Just with a simple prompt. Better than Gemini with SVG's. Google is cooked.\n>\n> A lot more coming soon. pic.twitter.com/nq0KPsIXwN\n>\n> — can (@marmaduke091) June 5, 2026\n\nRed-team access has typically preceded a wider rollout by a week or two, suggesting a plausible launch in the second half of June, close to when OpenAI's rumored GPT-5.6 (codename kindle-alpha) is also expected.\n\n> 🚨 NEW GPT-5.6 CHECKPOINTS DROPPED\n>\n>  OpenAI is testing 2 new checkpoints:\n>\n> > - kindle-alpha (release candidate)\n> > - kepler-alpha\n>\n> Join Our server we are serving back to back test before launch 🫣 pic.twitter.com/oalhzIV092\n>\n> — Chetaslua (@chetaslua) June 5, 2026\n\nEarly discussions, including creative venues like VoxelBench, suggest that Oceanus outputs are significantly superior to those of current models, even with minimal effort, though nothing has been independently confirmed. The leap looks promising, though not yet settled.\n\nOceanus on VoxelBench\n\nEarlier reports initially linked Mythos to Claude Code and Claude Security, targeting developers and security teams. Whether it will also reach business, personal, or Max tiers, or remain enterprise-only, is unclear; Pro almost certainly waits.\n\nJoin Dev Mode Discord for more!\n\n\n                            Join\n                        \n\nThis development coincides with Anthropic's new Institute paper, which argues that AI is already accelerating AI development, citing Mythos Preview achieving a 52x training-optimization speedup. Notably, the company frames this as a cautionary note, emphasizing that the field has not yet achieved recursive self-improvement and calling for verifiable methods to slow down, rather than celebrating prematurely.",
  "title": "Anthropic started red teaming new Mythos models, first results",
  "updatedAt": "2026-06-05T15:17:47.213Z"
}