Test-time scaling

Technique · Architecture & Efficiency · 3 items since May 2026

Spending more compute at inference, through longer reasoning, sampling, or verification, to get better answers.

Organizations: BAAI NVIDIA
Techniques: Context compaction Curriculum learning Reinforcement learning with verifiable rewards

Trends

Sep 11
NVIDIA's open recipe for IMO gold
A fully open, natural-language test-time-compute pipeline on Nemotron 3 Ultra, using no formal prover, external tools, or internet, scored 30/42 at IMO 2026 to meet the gold threshold, and ships with checkpoints, training data, code, and Nemotron-IMO-Bench.
Jul 24
BAAI's AREX self-improving research agent
A deep-research agent alternates an inner evidence-gathering loop with an outer constraint-wise self-audit and an autonomous context-compression tool, released as open-weight 4B and 122B-A10B MoE models competitive on BrowseComp and Humanity's Last Exam.
May 14
SU-01 reaches olympiad gold at 30B-A3B
A 30B-A3B MoE trained with reverse-perplexity curriculum SFT, coarse RLVR then proof-level RL, and test-time verification reaches gold on IMO 2025, USAMO 2026, and IPhO 2024–25, scoring 35/42 and matching the top human among 340 USAMO competitors.