An 11B-active model reporting 96.11% overall on MathArena and 97% on AIME 2026 I at about $0.40 per run sets a cheap reference point for math and reasoning workloads previously assigned to far larger models.

• postvLLM gains an experimental Attention-FFN disaggregation plugin for MoE
• postOpenRouter free tier for Step 3.5 Flash jumps to 10K calls a day
• article👏🏻Congratulations！Step3-VL-10B was selected for HuggingFace Daily Papers…
• postStep-Audio-R1.1 opens weights for a speech model that reasons in real time

• postGLM-5.3-Flash lands on the intelligence-per-dollar Pareto frontierArtificial Analysis
• postStep 3.5 Flash: 196B MoE open weights built for agent workloadsStepFun
• postStep 3.7 Flash: Apache 2.0 multimodal MoE tuned for agent efficiencyStepFun

Checking sign-in…

Loading comments…