Mancer: Weaver (alpha)

Mancer: Weaver (alpha), created on August 1, 2023, consistently ranks among the fastest models available and offers highly competitive pricing, demonstrating strong efficiency in both speed and cost across various benchmarks. Designed for roleplay and narrative situations, its performance reflects this specialized focus. While excelling in speed and affordability, the model exhibits significant limitations in accuracy across most evaluated categories. Its performance in Coding (3.0% accuracy), Instruction Following (4.0%), Email Classification (48.0%), and General Knowledge (0.5%) is notably low, often falling into the lowest percentiles. The model achieved 0.0% accuracy in the Ethics benchmark, indicating a complete lack of ethical reasoning in this test. Its strongest relative performance was in Reasoning (46.5% accuracy, 32nd percentile), suggesting some capacity for complex problem-solving, though still below average. The description notes an attempt to recreate Claude-style verbosity without the same level of coherence or memory, which aligns with the observed accuracy deficits in tasks requiring precise understanding and recall. Overall, Mancer: Weaver (alpha) is a cost-effective and fast option, but its utility is currently limited to applications where high accuracy and robust reasoning are not primary requirements.