Every question, answered once
470 questions, collected from every page on this site and grouped by what they are about. Each one links to the page that answers it in context, because a price is easier to trust next to the table it came from.
About the service
15What this is, what it is not, and the things worth knowing before moving production traffic to a smaller provider.
Cost and billing
64What a request costs, how the balance behaves, and what happens at the edges — a failed call, a cancelled stream, an exhausted balance.
Models
180Per-model questions: the rate, the licence, the context window, how to call it, and whether it is servable today.
Choosing for a workload
24Which model suits which shape of work, and why the answer changes with the input-to-output ratio rather than with the benchmark.
Model comparisons
113Head-to-head questions between two models in the catalogue, answered on price and specification rather than on quality claims we have not measured.
Switching provider
74Migration questions for people arriving from another provider, including the cases where staying is the better call.
The ones people ask first
Answered here in full. Everything else is one click away, on the page it belongs to.
- What is AI Token Router?
- AI Token Router is an OpenAI-compatible API gateway for open-weight AI models. Point your existing OpenAI SDK at one endpoint and you get Kimi K3, GLM-5.2, DeepSeek V4 and open-source video models like Wan 2.2 and LTX-2.5, at 36–42% below official pricing, rising to 65–68% on agent and chat workloads where a repeated prompt prefix is billed at the cached rate. One key, one bill, and the official rate printed next to ours on every row.
- How can you be 36–42% cheaper than official pricing?
- We buy inference capacity in bulk, run it at high utilisation, and pass most of that discount through instead of keeping it as margin. There is no platform fee and no surcharge on top of the per-token rate you see. It is also why we only carry open-weight models: their licenses permit third-party hosting, so we control the cost base rather than reselling someone else's API at a markup.
- Do I have to rewrite my code to switch?
- No. The API is OpenAI-compatible, so if you already use the OpenAI SDK -- or any framework built on it, which is most of them -- switching is a one-line change to your base URL. Request shapes, streaming frames, the usage block and the error envelope all match what your code already handles.
- Which models do you support?
- Twenty-four open-weight models are catalogued and seven are callable right now: Kimi K3, Kimi K2.6, GLM-5.2, DeepSeek V4 Pro, DeepSeek V4 Flash, Qwen-Image and Qwen3-Embedding-8B. The rest -- GLM-5.2 Air, Qwen3, MiniMax M2, Llama 4, Mistral Large 3, GPT-OSS, and the video models Wan 2.2, LTX-2.5, HunyuanVideo 1.5, CogVideoX and Mochi 1 -- are listed with published prices but are not yet served, and every model page and pricing row says which is which rather than letting you find out at call time. New open-weight releases are added at a published price as capacity for them comes online.
- How does cached-token pricing work?
- When you resend the same prefix -- a long system prompt, a fixed tool schema, a document you keep querying -- it is served from the model's KV cache and billed at the cached-input rate, typically 80-90% below the standard input rate. It happens automatically with no header to set, and a cache entry lives about five minutes after its last use. For agent loops this is usually the largest line on the bill, which is why the cached rate has its own column on the pricing table instead of a footnote.
- Are there hidden fees?
- No. The per-model rate is the whole price. No platform fee, no minimum spend, no card-processing surcharge at checkout, and no charge for failed requests. Worth checking when you compare: OpenRouter adds roughly 5.5% on credit purchases, which does not appear on their per-model pricing table.
- What stops a runaway agent from running up a huge bill?
- Your prepaid balance is a hard ceiling, enforced by the gateway on every request — a runaway agent stops at what you have already paid, never at an invoice. Unlike a configured cap, it cannot be misconfigured. Automatic recharge is off by default, and when enabled it is bounded too: a fixed amount, capped at three recharges in any 24 hours, switched off after three consecutive declines. Per-key limits are on the roadmap.
- What happens if a model provider has an outage or rate-limits me?
- The request fails with the upstream status and a machine-readable error code rather than being silently retried on a different model, and the failure is not billed. Automatic cross-model fallback is not implemented yet: substituting a model changes the output, the price and the licence terms of what you receive, so we would rather return an honest error your code can branch on than quietly hand you a different model's answer. Rate-limit responses come back as 429 with `rate_limit_exceeded`, which is distinct from the 429 with `insufficient_credits` you get when the balance runs out.
- How do you handle my prompts and data?
- We do not train on your prompts or completions, ever. Prompts and completions from a normal chat or completion request are relayed and not stored by us at all. The inputs and results of asynchronous jobs -- video and image generation, which cannot return inline -- are held for 30 days from completion so you can fetch them, or deleted immediately on request. Usage metadata (model, token counts, latency, cost) is kept for 24 months, because that is what the dashboard and your invoice are built from. The full schedule is in the privacy policy, and the numbers there are the authoritative ones.
- Can I bring my own provider API key?
- Not yet. Routing your own negotiated provider contract through us -- keeping your rate while gaining logs and per-key attribution -- is designed and on the roadmap, but it is not shipped, so today every request is billed against your prepaid balance at our published rate. We would rather say so here than let you discover it after signing up.
- How is this different from OpenRouter?
- OpenRouter is a router across 500+ models including closed frontier models like GPT, Claude and Gemini, with multi-provider failover. If you need those, OpenRouter is the better fit and we say so on our comparison page. We are narrower on purpose: open-weight models only — 7 callable today out of 24 catalogued, and the catalogue marks which is which rather than failing at call time — with no platform fee, published cached-input rates, and open video generation treated as a first-class category rather than something we do not carry.
- Why only open-weight models?
- Because reselling closed-weight model access frequently violates the origin provider's terms of service, and that is a risk we will not take on or expose customers to. Open licenses permit third-party hosting, which is what lets us run the inference ourselves, control the price, and publish it. It is also why we do not serve Sora 2, Veo 3.1, Runway Gen-4, Kling, or Wan 2.5/2.6 -- the last of which is API-only, with weights never released, despite the open Wan 2.1/2.2 lineage.
- How quickly can I make my first call?
- About twenty seconds. Sign up with GitHub, Google or an email and password, and your API key is shown on the dashboard immediately along with a copy-runnable snippet. New accounts get $5 in free credits, so the first calls cost nothing and no card is required. Every model page also carries a snippet with the model id already filled in.
- Do I need a credit card to start?
- No. Sign up with GitHub or Google -- no password to set, nothing to leak -- and you get $5 in credits plus 2M free tokens on any open-weight model. You only add a payment method when the free credit runs out, and there is no subscription -- it is pay as you go with no minimum spend, cancel anytime.
- Do you offer volume or enterprise pricing?
- Yes. Accounts spending $1,000+ per month qualify for custom rates, with deeper discounts, priority support and dedicated SLAs at higher tiers. We reply within 24 hours.
Still deciding?
Put your own volume into the calculator, or start with $5 of credit and check it against a real request.
No card required. The balance is a hard ceiling.