type: entity
created: Mon Apr 06 2026 02:00:00 GMT+0200 (Central European Summer Time)
updated: Mon Apr 06 2026 02:00:00 GMT+0200 (Central European Summer Time)
sources: raw/plans/openrouter-llm-research, raw/plans/invoice-processing-plan, raw/notes/reference_api_keys
tags: tool ai llm-api
aliases: OpenRouter API

OpenRouter

abstract

LLM API gateway providing access to 312+ models. Used in UnitCycle for invoice extraction via Gemini 2.0 Flash ($0.10/$0.40 per M tokens). OpenAI-compatible API.

Usage in UnitCycle

Primary model: google/gemini-2.0-flash-001 — fast, cheap ($0.10 input, $0.40 output per M tokens)
Cost per invoice: ~$0.0004
API key in backend/.env as OPENROUTER_API_KEY
Endpoint: https://openrouter.ai/api/v1/chat/completions (OpenAI-compatible)

Research Findings

Ranked comparison of 312+ models (from research doc):

Best for extraction: Gemini 2.0 Flash (best price/performance)
Alternatives: Qwen3-32B (free tier), DeepSeek V3.1 (good quality)

Sources

raw/plans/openrouter-llm-research — model comparison research
raw/plans/invoice-processing-plan — integration architecture
raw/notes/reference_api_keys — key storage

Related

wiki/concepts/invoice-processing — primary use case
wiki/entities/llamaparse — paired in the extraction pipeline
wiki/concepts/tech-stack — infrastructure