Google DeepMind · Chinchilla

Chinchilla 70B

READER SUMMARY

A 70-billion-parameter compute-optimal language model trained on 1.4 trillion tokens.

DOOM SCORE34.1out of 100model risk profile, not the overall index

CURRENT ASSESSMENT · REVISION 1

Why this model scores 34.1

Strong language capability and influential scaling results raised frontier efficiency, while research-only access, little inherent agency, and minimal direct deployment kept the remaining dimensions moderate or low.

Capability55

Autonomy5

Deployment8

Misuse potential30

Control difficulty22

0 comments · 0 votesOpen discussion

No comments yet. Start the discussion.

MODEL-ATTRIBUTED EVIDENCE

News tied to Chinchilla 70B

The model score of 34.1 rates this model's risk profile. The overall Doom Index of 60.4 measures the recalibrated complete evidence corpus. These values answer different questions.

NET MODEL-ATTRIBUTED INDEX CONTRIBUTION+0.03

Each article's leave-one-out 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 are recalculated from their technical profile and related news, but never feed back into the overall index.

29 Mar 2022CapabilityTOWARD

DeepMind shows 70B Chinchilla outperforms much larger language models

DeepMind's compute-optimal scaling study trained Chinchilla on far more data and showed the 70-billion-parameter model outperforming substantially larger systems at the same compute budget.

Full item contribution: +0.03
Chinchilla 70B equal share: +0.03

Read assessment →

AUDIT TRAIL

Model score history

R1
Doom Score 34.1
New exact public research model with original timestamp and five evidence-based dimension inputs.
12 Aug 2026