Case StudyApr 15, 2026
The Hidden Cost of Using GPT-4 for Everything
Most teams default to the most capable model. Here's why that's leaving money on the table — and what the data says about task-appropriate model selection.
Read More →
BenchmarksApr 3, 2026
Benchmark: Flux vs. Manual Model Selection
A walkthrough of our offline, synthetic routing benchmark: up to 89% estimated savings in our benchmarks — 84.8% vs GPT-4o, 89.3% vs Claude Sonnet 4.6, 82.2% vs Gemini 2.5 Pro — and what these numbers do and don't tell you.
Read More →
EngineeringMar 22, 2026
Per-Customer Tuning: Personalizing LLM Routing at Scale
How customer_id lets you scope adaptive learning and cost ceilings per end customer — stricter limits for free-tier traffic, higher quality floors for premium.
Read More →
EngineeringMar 10, 2026
Adaptive Learning: How Flux Gets Smarter Over Time
How the per-(model, task) exponential moving average (EMA) updates from every response, so routing sharpens over time once you persist adaptive state.
Read More →
Open SourceFeb 28, 2026
Why Open Source? The Flux Philosophy
Transparency, community trust, and why we chose AGPL-3.0 licensing for Flux — and why that's better for you than a black-box routing service.
Read More →
ProductFeb 15, 2026
Cost Calculator: How Much Can You Save?
An interactive walk-through of the savings model: how we calculate the estimate, and what factors — workload, provider pricing, configuration — change it in practice.
Read More →