Introducing Aiven AI Gateway

At home I run fully automated coding loops on a few side projects. They use more than 100-million tokens a day. At Claude Opus 5.5 prices, that costs more than $5,300 a month.

For hobby projects.

The same usage on DeepSeek V4.1 Flash cost about $330 a month, one sixteenth of the price. If we want to keep scaling how much we use AI, most of those tokens have to run on open models, period.

Most engineers, including at Aiven, still work next to an agent: they prompt it, review the work and prompt again. That’s still a primitive use of the technology - the next step is unsupervised agentic engineering that runs all day. This future is inevitably coming, the only question is when does the user and their organization catch up to always-on Inference.

And when they do catch up, they will find that five hundred of those loops cost more than $2.6 million a month on Opus 5.5, and that's before the rest of the business automates anything. On DeepSeek V4.1 Flash, these cost about $165,000.

This is why we built Aiven AI Gateway. It's a fully open source implementation that gives us models from multiple providers with a particular focus on open source.

Built initially for our internal use, today we are making it available to all our customers.
Aiven customers with coding agents need no code changes at all:

Loading code...

While Getting started with the AI Gateway is one line:

Loading code...

Open models stopped being the cheap, worse option

For the past year, open models like DeepSeek sat at the bottom of the pyramid: the low-cost fallback when you couldn't afford the best. That is no longer true. In OpenRouter's September snapshot, open models handled 72% of token volume.

They aren't winning on price alone. They win the way Kafka and PostgreSQL won: the open ecosystem’s network effect. One team makes the model faster. Another makes it smaller. Providers compete to serve it. Developers bend it to their workflows. Every new application gives someone a reason to improve it, and every improvement makes the next application easier to build.

Pricing

During early access, usage is free and nothing is invoiced. Existing Aiven customers can request access for your organization, and once we approved the organization gets $300 in credit. New organizations can request access too, and we approve them as capacity allows.
$300 goes a long way on open models. An agent loop that uses 100 million tokens a day on a comparable open-source model costs about $11 a day, so the credit lasts about four weeks. At Claude Opus 5.5 prices, the same loop would use it up in under two days.

What's coming next

Aiven is investing heavily in AI. This gateway is simply the first step in something big.

• Routing: named routers, with fallbacks across models and providers
• Pay-as-you-go billing
• Budgets and spend alerts that you set
• Bring-your-own-key support