Research
One of my favorite things about Sakana Fugu is the recursive test-time scaling. When allowed to call itself recursively, it reads its own pr…
One of my favorite things about Sakana Fugu is the recursive test-time scaling. When allowed to call itself recursively, it reads its own prior output and spins up corrective workflows on the fly. We
One of my favorite things about Sakana Fugu is the recursive test-time scaling. When allowed to call itself recursively, it reads its own prior output and spins up corrective workflows on the fly. We are opening up the API for beta testers to try it out: https://sakana.ai/fugu-beta/
Related
- When More Thinking Hurts: Overthinking in LLM Test-Time Compute Scaling
- Stability and Generalization in Looped Transformers
- C-voting: Confidence-Based Test-Time Voting without Explicit Energy Functions
- Optimal Aggregation of LLM and PRM Signals for Efficient Test-Time Scaling
Source: David Ha (X) | 2026-04-24