Otari Subscriptions: Making open-weight models the easy choice for coding
A year ago, if you wanted a model that could work through a real codebase, call tools, and finish a multi-step task on its own, you reached for a closed model in a closed harness. That's no longer true.
Open-weight models have caught up
Models like GLM-5.3, DeepSeek V4 Pro, Kimi K3 and Qwen3.8 can now handle the work coding agents depend on: planning a change, editing across files, running tools, and recovering when something breaks. Each new release narrows the gap, and the pace isn't slowing.
Because their weights are open, anyone can host them, fine-tune them, or switch providers without asking permission. That's why more teams are moving their coding work to open-weight models: lower cost, more control, and no lock-in to a single vendor.
Why adopting them still feels hard
The models are ready, but using them day to day isn't simple yet. There are too many choices: which model, and which provider serves it well? Every provider means another account, API key and bill, plus configuring your coding tool for each one. Closed-model plans bundle web search and code execution, while open models usually come bare. And per-token pricing makes a month of coding hard to budget.
That’s why many developers stick with the default, even when an open model would do the job just as well.
What Otari Subscriptions does
One monthly plan removes that friction:
- The best open-weight models, one key, with new models added as they're released.
- Routed for price and prompt caching. Otari sends each request to the provider that serves the model best on price, health and latency, and keeps your session on the same provider so prompt caching keeps working.
- Works where you already build. Pi, Hermes, Claude Code, Codex and OpenCode. Point your harness at Otari and that’s it.
- Server tools included. Web search, web fetch and code execution come with every plan, with a monthly allowance.
- A predictable monthly price. No service fee on your subscription usage. If you go past your plan, usage continues from your organization's balance, so your work doesn't stop.
- Control built in. Routing, budgets and guardrails come with every plan.
- Built on open source. The Otari gateway is open source, so you can run it yourself and you're never locked in to us either.
Plans (early access pricing)
- Otari Go: $10 a month for $30 of usage
- Otari Pro: $40 a month for $120 of usage
Plans are personal and cover open-weight models. Closed models such as Claude and GPT are paid from your organization's balance, as they are today.
Early access
We're opening with limited spots and sending invitations in waves over the coming weeks. A Teams plan is also coming soon.
At Mozilla AI, we think open models should be the default choice for developers, not the hard one. Otari Subscriptions is our step toward that.
Request early access at otari.ai/subscriptions
