System record

AlphaEvolve / Gemini ensemble

Organization: Google DeepMind · base model: Gemini Flash, Gemini Pro

System and setup

Reported system information
Model versionnot reported
Software used to run the modelEvolutionary program search
Software versionnot reported
Tools available to the modelnot reported
Release dateNot reported

Autonomy assessments

source author assessmentadjacent

ai rd frontier model development

Framework levels: B0 · assessed 2026-09-10

Assessor: Duan et al.

Duan et al. place the task-specific AlphaEvolve program-search boundary at B0 in their industry table. Deployment of generated artifacts is a different boundary from autonomous inheritance by the improver.

Decisions or controls outside the AI system: Task-specific evaluators; Human production integration

Evidence against this assessment: None cited in this assessment

Extraction: agent checked. This checks whether the record matches its source; it does not mean the experiment was repeated independently.

Recursive-improvement evidence

tracker evidence synthesis

Assessment by RSI Tracker (Codex evidence synthesis) · 2026-09-26 · extraction review agent checked.

Resource budget:unknown · See study resource accounting; evaluation constraints are distinct from cumulative search cost.·Independence:known · Source-author report; no independent replication recorded.

Retention

not reported

Not recorded in this dataset.

Transfer

not reported

Not recorded in this dataset.

Stability

not reported

Not recorded in this dataset.

Meta-recursion

not reported

Not recorded in this dataset.

Published capability results

Benchmark and indicator evidence
MeasureVersionReported resultDate
AlphaEvolve training optimizationMay 2025 report1%2025-05-14
AlphaEvolve training optimizationMay 2025 report23%2025-05-14

Related research

AlphaEvolve: iterative artifact optimizationreport · 2025-05-14Primary source status belongs to the linked paper; the tracker assessment is separate.
Additional system details
System ID
alphaevolve
Organization
Google DeepMind
Base model
Gemini Flash, Gemini Pro
Release date
Not reported
Assessment records
1
Recursive evidence records
1