Can I call any video model on this platform today?

No. Wan 2.2 T2V, Wan 2.2 I2V, LTX-2.5, HunyuanVideo 1.5, CogVideoX-5B and Mochi 1 are all catalogued at a real, published rate, and none currently have a configured upstream serving them. A request naming one returns a clear error.

Which image model can I actually call today?

Qwen-Image. It is the only one of the four catalogued image models — the others being FLUX.2 [schnell], Stable Diffusion 3.5 Large and HiDream-I1 — currently served by a configured upstream.

Why publish a price for a model you can't serve?

So the catalogue stays complete and comparable rather than silently shrinking to only what happens to be live this week. Every unservable model is marked plainly, and the price shown is what it will cost once served, computed the same way as every callable model's rate.

Resources · AI Token Router

PricingMedia modelsOpen-weight

Video and Image Generation Models: What's Priced and What's Actually Callable

Of the video models in this catalogue — Wan 2.2 T2V, Wan 2.2 I2V, LTX-2.5, HunyuanVideo 1.5, CogVideoX-5B and Mochi 1 — none are currently callable through this platform, and of the image models, only Qwen-Image is. Every one of them still carries a real, published rate, computed against its publisher's own price the same way every other model on this site is: the pricing is set and it is real, but the availability is not there yet for most of the lineup, and we are stating that plainly rather than letting a rate card imply a channel that does not exist.

September 8, 2026

Put your own numbers in before you take ours on trust.

See the full video and image lineup

What's catalogued vs what's callable

This site catalogues models at their intended rate whether or not a configured upstream currently serves them, and says so clearly rather than removing an unservable model and re-adding it later once it becomes available. The video and image lineup is where that policy matters most: six video models are catalogued and none are callable today; four image models are catalogued and only one, Qwen-Image, is.

A request naming an unservable model returns a clear error rather than a fabricated result. Nothing below is a promise about when that changes — it is the current, checkable state of the catalogue.

The video lineup

None of the six video models below can currently be called through this platform. The rates shown are the intended prices, ours against each publisher's own, computed the same way as every callable model's rate — they are what these models will cost once served, not a rate you can bill against today.

Video input rate, our price against official — none currently callable

Every model's intended rate here already undercuts its publisher's own price by a real margin, which says nothing about when any of them will actually be servable. On input, CogVideoX-5B is the lowest at $0.017 per second and Wan 2.2 I2V A14B the highest at $0.032 per second, a 1.9x spread. Every rate here runs 40–43% below the model publisher's own.Our published rates, read from the same table the API bills from.

The image lineup

Image is a different story: one of the four catalogued models, Qwen-Image, is callable today. The other three are priced but not currently servable, on the same basis as the video lineup above.

Image rate, our price against official — one of four currently callable

Qwen-Image is the one bar in this set you can actually call today; the other three sit at their intended rate with no upstream currently serving them. On input, FLUX.2 [schnell] is the lowest at $0.0018 per image and Stable Diffusion 3.5 Large the highest at $0.0210 per image, a 11.7x spread. That is 40% below the model publisher's own rate.Our published rates, read from the same table the API bills from.

Why we publish a rate for a model we can't yet serve

The alternative is to leave a catalogued model off the site entirely until it is servable, which trades one kind of honesty for another: a visitor comparing this catalogue against a competitor's would see a shorter list than the one actually being tracked, and would have no way to tell a temporary gap from a permanent decision not to carry a model. Marking a model's availability clearly and leaving its intended price visible lets a visitor plan around what is actually true today without losing the information for when that changes.

This is the same policy the rest of the catalogue runs on — seventeen of the twenty-four models here are not currently callable, and every one of them is marked, not hidden.

Questions this raises

Can I call any video model on this platform today?: No. Wan 2.2 T2V, Wan 2.2 I2V, LTX-2.5, HunyuanVideo 1.5, CogVideoX-5B and Mochi 1 are all catalogued at a real, published rate, and none currently have a configured upstream serving them. A request naming one returns a clear error.
Which image model can I actually call today?: Qwen-Image. It is the only one of the four catalogued image models — the others being FLUX.2 [schnell], Stable Diffusion 3.5 Large and HiDream-I1 — currently served by a configured upstream.
Why publish a price for a model you can't serve?: So the catalogue stays complete and comparable rather than silently shrinking to only what happens to be live this week. Every unservable model is marked plainly, and the price shown is what it will cost once served, computed the same way as every callable model's rate.

AI Token Router is an OpenAI-compatible gateway for open-weight models, priced below each publisher’s own rate on every row.

See the full video and image lineup Get $5 in free credits

What a Fixed Monthly AI Budget Actually Buys in 2026
Three realistic budget tiers, worked by hand against this catalogue's own rate table, at one stated request shape -- how many requests and tokens $10, $50 and $250 a month actually buys on three callable models.
When a Closed Frontier Model Is Still the Right Call
Closed frontier models measurably lead reasoning-heavy benchmarks as of September 2026. Where that lead and a simpler operational model are worth the higher price -- and why our catalogue is not the answer for that reader.
Self-Hosting vs a Managed Open-Weight API: When Each Wins
Where the self-host breakeven actually sits, what self-hosting really costs once engineering time is priced in, and the honest cases where self-hosting wins -- this is not a blanket argument for a managed API.