Comments
>log in to comment
Hi DevHunt! I'm Joan-Marc, founder of BiVelio. We built Savings to help developers cut LLM API costs and understand what each optimization saves. The engine is free and open source (Apache-2.0). Run the SDK on your own infrastructure with your provider's API key. It combines caching, request deduplication and prompt optimization. Each request returns a SavingsReport that separates measured savings from estimates. Your prompts never pass through BiVelio's servers. Optional Savings Cloud adds a team dashboard. Savings depend on your workload. We'd love feedback on integration, caching behavior and the clarity of the reports. What would you need to verify before using this in production? Try it on a representative workload and share what works, what breaks and where it doesn't save. I'll be here to answer questions.
More AI tools →80 free dev tools →
BiVelio Savings Layer is an open-source SDK that caches, deduplicates and compacts LLM requests to cut API costs.
- for
- Developers building Node.js/TS apps or agents that call LLM APIs.
- pricing
- freemium
- license
- Apache-2.0
- v1.2.1BV-SALA 1.2.1 — README revisado para la comunidad + paper v2.3-r16 days ago
- py-v1.2.1Python client 1.2.1 — alineado con 1.2.16 days ago
- v1.2.0BV-SALA 1.2.0 — cache-aware prompt management + cross-provider shadow + live evi7 days ago
Key features
- Request deduplication — Repeated identical calls are served from memory without contacting the provider.
- Prompt token compaction — Compresses prompts to send fewer tokens, reducing input cost.
- Per-request SavingsReport — Each call returns a report showing measured vs estimated savings.
- Zero-telemetry open core — The SDK runs locally with no default data collection or required account.
- Optional cloud dashboard — Savings Cloud provides a team dashboard, shadow-mode measurements and flat-fee seat licensing.
- Multi-provider support — Works with OpenAI, Anthropic, Gemini, Mistral, DeepSeek, Azure OpenAI, vLLM, Ollama and any OpenAI-compatible endpoint.
Use cases
- Cache repeated LLM calls in a chatbot to avoid extra provider charges.
- Compress long prompts in a code-generation pipeline to lower token usage.
- Measure potential savings across a team’s LLM usage before enabling optimizations.
BiVelio Savings Layer pricing
- MeasureFreeShadow mode only · Up to 5 seats · 20,000 requests/day
- Pro€15 per seat per month/monthAll optimizations enabled · Per-request receipt · Flat fee, no commission
- Team€129 per month/month10 seats included · Additional seats €12/month · Flat fee, no commission
BiVelio Savings Layer vs alternatives
BiVelio Savings Layer | supermemory | Elasticsearch | |||
|---|---|---|---|---|---|
| Best for | Cost-saving LLM request layer | Agent memory storage | Agent state persistence | Search indexing | Search and analytics |
| Pricing | Freemium | Subscription | Subscription | Subscription | Subscription |
| DevHunt upvotes | 1 | 0 | 0 | 0 | 0 |
| Launched | Nov 2026 | — | — | — | — |
- BiVelio Savings Layer vs Mem0: Mem0 provides a memory layer for agents, while BiVelio focuses on cost-saving via caching and token compaction.
- BiVelio Savings Layer vs supermemory: supermemory offers persistent memory for agents; BiVelio targets API cost reduction rather than state storage.
- BiVelio Savings Layer vs Algolia: Algolia is a search API, unrelated to LLM request optimization that BiVelio provides.
- BiVelio Savings Layer vs Elasticsearch: Elasticsearch is a distributed search engine, not a cost-saving SDK for LLM calls.
BiVelio Savings Layer FAQ
Do I need a BiVelio account to use the SDK?+
No. The open-source engine runs on your infrastructure without an account or API key.
Will my prompts be sent to BiVelio servers?+
Only the optional Savings Cloud receives counts and costs; prompt content never leaves your infrastructure.
What is the difference between Community and Measure?+
Community is the free local engine; Measure is the free cloud plan that shows shadow-mode savings in a dashboard.
Can I keep using my existing LLM provider?+
Yes. You configure the provider and API key; the layer works with any OpenAI-compatible endpoint.
How are seats counted?+
A seat is a machine or service identity that runs the layer; CI or ephemeral runners are excluded.
Is there any commission on the savings I achieve?+
No. All paid plans are flat-fee per seat with zero commission on savings.
Summarized by DevHunt from savings.bivelio.com · Nov 20, 2026. Details may change; check the official site.

BiVelio Savings Layer
supermemory
Elasticsearch







