AI reasoning models shift the frontier as open weights challenge closed labsA pivot towards test-time compute and open-weight reasoning architectures is narrowing the gap between proprietary frontier labs and accessible developer models.11 hours ago
DeepSeek and open-source reasoning models challenge OpenAI in frontier AI performanceOpen-weights AI systems using chain-of-thought reinforcement learning are matching proprietary benchmarks at a fraction of the cost.10 October 2026
