• 2 min read
Ramp opens its LLM router after 30% cost cuts
Ramp says its new Router cut internal LLM costs by about 30% while handling 2.75T+ tokens a month with 99.999% routing reliability.

Image: Hacker News
Ramp is opening up Ramp Router, an API endpoint that routes requests across multiple model providers based on quality, cost, and availability. The company says it built the product to manage 100+ AI use cases internally, and that Router cut its LLM costs by ~30% while making features “smarter and faster.”
According to Ramp, the service now routes 2.75T+ tokens monthly, with 99.999% successful routing and about 30 ms of added latency.
The pitch is straightforward: instead of wiring an application to a single model provider, developers send requests to Ramp Router, which can automatically move eligible traffic to a cheaper model if it still meets the required quality bar. Ramp says the system continuously tests new models on real workloads as the “price-intelligence-latency frontier” shifts.

Recommended reading
AI’s next bottleneck may be materials, not chips
Ramp highlights a wide mix of workloads for the product, including:
- invoice line-item extraction
- support ticket classification
- board deck summarization
- contract risk review
- transaction anomaly explanation
- SQL generation
- API debugging
- document translation
- sales email drafting
- user-message moderation
The company lists support for models from OpenAI, Anthropic, and Gemini, plus select open-source models including Kimi. On the page, Ramp also references evaluated or selected models such as Gemini 3 Flash, Claude Haiku 4.5, Claude Sonnet 5, Gemini 3.5 Flash, Claude Opus 4.8, Grok 4.5, Claude Opus 4.6, Claude Fable 5, Gemini 3.1 Pro, and Gemini 3.1 Flash Lite.
“At Ramp, Router cut our LLM costs by 30% while making our features smarter and faster.”
Ramp says the product uses an OpenAI-compatible API, so existing integrations should require only a base-URL change. It also bundles caching, compaction, semantic attribution, and 100+ optimizations per request, alongside spend tracking by model, product, team, and project through Ramp Token Spend Management.
At launch, Ramp Router is free to use. Users pay list price for tokens, and the first 500 people invited will get $100 in promotional credits. Ramp says early access does not require being a Ramp customer, having a company account, or even an LLC.
Enterprise Editor
Marcus follows the money. He covers enterprise software, cloud architecture, and the tectonic shifts in Big Tech strategy. He translates dense earnings calls and complex M&A activity into actionable insights about where the industry is actually heading. If a tech giant makes a silent pivot, Marcus is usually the first to notice.
via Hacker News


