DOOMBENCH.COM
OpenAI 🇺🇸 · GPT-Rosalind

GPT-Rosalind-6

Trusted-access life-science deployment of GPT-6 Astra weights. OpenAI treats biological capability as High and cybersecurity as Critical, while none of four Critical biological evaluations met its indicative threshold. Additional customer screening, human review, least privilege and cyber blocking apply. Artifact licensing remains unverified in this review.

DOOM SCORE85.5out of 100model risk profile, not the overall index
0 comments · 0 votesOpen discussion

Public discussion is readable by everyone. Sign in to comment, reply, or vote.

No comments yet. Start the discussion.

Send detailed or private feedback →
CURRENT ASSESSMENT · REVISION 1

Why this model scores 85.5

Capability 100 and autonomy 98 inherit the exact Astra weights and its broad agent profile, not the narrow application label. Deployment 25 reflects approved research customers and enterprise controls rather than general access. Misuse 100 reflects Critical cyber capability and fewer dual-use biology refusals. Residual control difficulty 85 remains high despite layered screening, monitoring and blocking; tool tests show warning circumvention, not an actual escape. Availability stays Unknown because proprietary licensing was not established by accessible sources.

Capability100
Autonomy98
Deployment25
Misuse potential100
Control difficulty85
MODEL-ATTRIBUTED EVIDENCE

News tied to GPT-Rosalind-6

The model score of 85.5 rates this model's risk profile. The overall Doom Index of 69.1 measures the complete temporally weighted evidence record. These values answer different questions.

NET MODEL-ATTRIBUTED INDEX CONTRIBUTION+0.26

Each article's current Doom Index contribution is divided equally among the exact models named on that article. This prevents multi-model evidence from being claimed in full on several model pages. Model risk scores use a bounded temporal offset around their technical profile, but never feed back into the overall index.

Deployment TOWARD

OpenAI deploys Rosalind-6 with trusted biological access and Critical cyber capability

OpenAI reports that GPT-Rosalind-6 is deployed to approved life-science customers using GPT-6 Astra weights, with fewer dual-use biological refusals. It remains High, not Critical, in biological capability and inherits a Critical cybersecurity assessment. None of four Critical biological tests crossed its indicative threshold. Access reviews, least privilege, human review and cyber blocking constrain deployment; tests do not establish real-world harm.

Full item contribution
+0.26
GPT-Rosalind-6 equal share
+0.26
Read assessment →
AUDIT TRAIL

Model score history

  1. R1
    Doom Score 85.5

    Creates the exact deployed Rosalind-6 variant from its dated October 30 system-card addendum; not a renamed Astra record or a claim of new weights.

    31 Oct 2026

Share this page

DoomBench social sharing card for GPT-Rosalind-6.

  1. GPT-Rosalind-6 by OpenAI has a DoomBench model risk score of 85.5 out of 100, based on five transparent version-specific dimensions rather than the overall index.

  2. GPT-Rosalind-6's highest current DoomBench dimension is capability at 100.0 out of 100; the profile publishes every component score and its editorial rationale.

  3. DoomBench links 1 source-backed evidence item to GPT-Rosalind-6, while keeping the model's risk profile separate from each item's contribution to the live Doom Index.

    https://www.doombench.com/models/openai-gpt-rosalind-6