Architecture Overview
Meida operates as an edge-distributed routing gateway positioned between your application and multiple foundation LLMs. Every turn passes through a 3-step pipeline before reaching a model.
Routing Pipeline
Semantic Classification
Each incoming turn is analyzed in under 4ms by a lightweight local classifier. It extracts intent, complexity score (0–1), context length, and risk flags — without ever touching a foundation model.
Tier Selection & Cost Arbitrage
Based on the classification output, the arbitrage engine selects the lowest-cost provider tier that can reliably complete the task. Simple queries go to fast/cheap models; complex reasoning escalates to premium tiers.
Zero-Downtime Failover
If the primary provider returns a 429 (rate limit), 503 (outage), or times out, the failover engine instantly reroutes to the next-best provider in the same tier. No tokens are lost, no latency spikes — the user never notices.
Edge Node Pool
The following providers are actively connected to the Meida routing mesh. Each node is health-checked every 5 seconds.
| Provider | Tier | Role | Status |
|---|---|---|---|
| Claude 3.5 Sonnet | T4 (Premium) | Complex reasoning, audits, long-context | PRIMARY |
| GPT-4o | T3 (Standard) | General purpose, balanced cost/quality | ACTIVE |
| DeepSeek-V3 | T2 (Fast) | Code generation, structured output | HOT |
| Llama-3 70B | T1 (Economy) | Simple queries, classification, chat | HOT |
Classify (4ms) + Arbitrage (8ms) + Network Hop (~30ms) = sub-50ms before first token.