Tag
Recent ternary 1.58-bit LLM releases from small labs demonstrate speed and medical specialization but struggle with long-horizon tasks, with optimism for future models to compete with larger architectures like Qwen.
StepFun's Step 3.7 Flash, a 198B sparse MoE model with 11B active parameters, matches 97% of Claude Opus 4.6's coding performance on SWE-Bench Verified at roughly one-ninth the cost, using an Advisor Mode strategy that reserves expensive frontier model calls for critical decision points.