Tag: benchmarks - Articles - Wain Blog

Do Speech Recognition Models Memorize the Test Answers? Three Probes Across 11 ASR Models

Do Speech Recognition Models Memorize the Test Answers? Three Probes Across 11 ASR Models

NVIDIA AVO Scores 100.00 on the ARC-AGI-3 Public Set by Changing the Harness, Not the Model

Anthropic Aug 22, 2026

NVIDIA AVO Scores 100.00 on the ARC-AGI-3 Public Set by Changing the Harness, Not the Model

Reproducing 2,226 ICML 2026 Papers with Agents - 23% Had a Claim Falsified, 242 Split the Verdict

ICML Analysis Aug 17, 2026

Reproducing 2,226 ICML 2026 Papers with Agents - 23% Had a Claim Falsified, 242 Split the Verdict

OpenAI Says Two API Settings Roughly Tripled Its ARC-AGI-3 Score