# Sakana AI — Recursive Self-Improvement through Multi-Agent Self-Supervision

- Company: Sakana AI (sakana.ai)
- Announced: 2026-10-11T15:00:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://sakana.ai/mass
- Record: https://forck.live/items/19304-recursive-self-improvement-through-multi-agent-self-supervision
- Subject: Sakana models

Sakana AI introduced Multi-Agent Self-Supervision (MASS), a method where a shared model uses a team of virtual subagents to solve tasks, searches for better multi-agent workflows, and selects the best ones using its own judgments. Training on the best team's executions distills collective experience back into the shared model. In two cycles using a 27B open-weights model on synthetic open-ended research tasks, the score per output token reached 1.2-1.6x the base model's level across four research benchmarks.

## Evidence

Verbatim from https://sakana.ai/mass:

> We introduce Multi-Agent Self-Supervision (MASS). One shared model solves tasks through a team of virtual subagents, searches for better multi-agent workflows, then uses its own judgments to select the best ones. Training on the best team's executions distills the team's collective experience back into the shared model.

---

Record: https://forck.live/items/19304-recursive-self-improvement-through-multi-agent-self-supervision
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md