Show HN: Echo – Fable-level results at 1/3 the cost using open-weight models
Points and comments are a snapshot, not live.
Echo routes prompts across open-weight models to match Fable results at lower cost.
Echo is an AI system that allocates computation across a pool of open-weight models (GLM-5.2, Kimi K2.7, and others) per request rather than using a single model for every task. Its evaluation shows consistent performance above the best individual model in its pool and aggregate results comparable to Fable at roughly one third the inference cost. The builder notes that weaker models can still be useful on specific problems. A chat interface and OpenAI-compatible API are available. The builder is studying failure cases and testing on coding and agentic tasks.
What commenters are saying
Commenters were broadly skeptical. Several compared Echo to existing model-routing systems like OpenRouter Fusion, Sakana Fugu, and NotDiamond, questioning its novelty. The decision not to disclose per-request routing decisions drew criticism: commenters argued that transparency is essential for debugging, observability, and legal compliance in production use. The builder responded that the routing policy is the product but acknowledged they are open to sharing aggregate allocation info. A minority defended the approach as different from simple model routing, focusing on allocation and combination rather than just selection.