Hailuo 2.3 API and Pricing, Explained Through a Real Pipeline
Hailuo 2.3 is the MiniMax video model before Hailuo 3, and on Treza it is the one with a single flat price: 8.17 cents a second, 6 or 10 second clips, 16:9 only. Here is the rate card, what our test render billed and returned, how it compares with Hailuo 3 Max for the same money, and how to reach it as a versioned API endpoint.
Hailuo 2.3 is MiniMax's video model from before Hailuo 3. It renders a 6 or 10 second clip in 16:9 from a text prompt, or from a still you hand it as the first frame, and OpenRouter lists one price for it at every setting: 8.17 cents per second. It has had its own page on Treza since October 14, 2026, it sits in the Video Generation node's picker beside Hailuo 3 Max, and you can ask for it by name in chat.
This post covers what Hailuo 2.3 costs, what our test render billed and came back as, how it compares with Hailuo 3 Max, what the endpoint accepts and leaves out, and how to call it as a published API endpoint. Every parameter and price below is read off Treza's live video model catalog as of October 2026, and the behavior described was checked with a real render.
What you get before you talk about price
| Capability | Hailuo 2.3 |
|---|---|
| Clip length | 6 or 10 seconds |
| Resolution | One setting, listed as 1080p; our render came back at 1366 by 768 |
| Aspect ratios | 16:9 only |
| Audio | None; our render carried no audio track |
| Keyframes | First frame only |
| Reference images | No, not on this endpoint |
This is a narrow model, and that is the point of it: two lengths, one shape, one price, no resolution dropdown to compare and no audio switch to forget. Capability detail lives on the Hailuo 2.3 model page.
How Hailuo 2.3 bills
Per second of output, at one rate:
| Setting | Published cost per second |
|---|---|
| Every duration, from a prompt or a first frame | $0.0817 |
So a 6 second clip is 49 cents of provider cost and a 10 second clip is 82 cents. A first frame does not change the rate, and there is no higher tier to drift into.
For scale, HeyGen Video starts at 2 cents a second at 480p, Veo 3.1 Lite at 3 cents for a silent clip at 720p, Wan 3.0 and Grok Imagine at 5 cents, and Hailuo 3 Max at 5 cents at 480p and 8 cents at 768p. Hailuo 2.3 sits at the top of that group, with less control to show for it.
What our test render billed, and what it came back as
We rendered one 6 second clip from a prompt on October 14, 2026: a barista pouring steamed milk into a latte in a sunlit cafe, slow push-in. Four things are worth knowing from it.
The bill matched the arithmetic exactly: six seconds at 8.17 cents is $0.4902, and that is what the provider charged. No minimum length, no rounding up.
The frame was not 1080p. The catalog lists 1080p as the endpoint's only resolution and exposes no resolution setting, but the file came back at 1366 by 768 pixels at 24 frames per second, running a hair under 6 seconds. Until a render shows otherwise, plan on 768p-class output. For a clip shown large, the AI video upscaler takes it up from there.
It was slow for its length. The clip took 71 seconds to come back. Hailuo 3 Max, the same vendor's newer endpoint, returned 5 second clips in 12 to 14 seconds in our launch-week tests.
And despite a prompt that asked for natural ambient sound, the file had no audio track at all, which matches the catalog's silent flag.
Hailuo 2.3 vs Hailuo 3 Max: the same money, with less control
Same vendor, nearly the same price at the default settings, and the two models split on almost everything else.
| Hailuo 2.3 | Hailuo 3 Max | |
|---|---|---|
| Price per second | $0.0817 at every setting | $0.05 at 480p, $0.08 at 768p |
| Clip length | 6 or 10 seconds | 5 to 15 seconds, every whole second |
| Aspect ratios | 16:9 | 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 |
| Keyframes | First frame | First frame, or first and last |
| Resolution setting | None; came back 1366 by 768 | 480p or 768p |
| Audio | None | No setting; quiet ambient track |
| Render time in our tests | 71 seconds for 6 seconds | 12 to 14 seconds for 5 seconds |
For most new work, Hailuo 3 Max is the better pick from the same vendor: a last frame, vertical and square shapes, a 480p draft tier at five eighths of the price, and far faster renders. The Hailuo 3 Max pricing post covers it in the same detail as this one.
Hailuo 2.3 earns its place in two cases: a channel or client library already rendered on it, where the next clip has to match and consistency is worth more than the controls you give up, and the 10 second widescreen clip at a known price, where 82 cents for a quiet establishing shot fits a budget without a tier to check.
If the soundtrack has to come from the model itself, neither is the answer. Hailuo 3 generates audio with every clip.
First frame: what it does, and what it does not
A first frame pins the opening image exactly, and the clip animates out of it; the first frame post walks through the technique and the image to video generator covers it from the still's side. It bills at the plain rate.
There is no last frame, so a shot cannot be pinned at both ends; Hailuo 3 Max takes both keyframes. And there are no reference images on this endpoint: a still wired into the node's reference input is used as the opening frame instead, not carried as a subject into a new scene. For that, use a model with reference input, such as HeyGen Video or Wan 2.7.
Because the model is 16:9 only, compose the first frame wide; a vertical still will not produce a vertical clip.
About the audio
Hailuo 2.3 renders picture only. Add sound in the graph: a Text to Speech node handles narration, an Audio Generation node handles music or effects, and the Sequence node mixes them over the picture. For a narrated piece, silent clips are an advantage rather than a gap, because there is no model-generated ambience to duck under the voice.
What you actually pay on Treza
Hailuo 2.3 through a Treza pipeline is metered from a prepaid credit balance. One credit equals $0.01, credit packs start at $5, credits never expire, and a failed generation charges nothing.
Before a run starts, the platform estimates the media cost of the graph and refuses to start one your balance cannot cover. For Hailuo 2.3 the estimator holds 9 cents of provider cost per second, a little above the 8.17 cent list rate, so a 6 second clip is held at 54 cents and settles at 49, with the difference returned to the balance.
For a finished video costed end to end, narration and captions included, see the AI video cost breakdown.
Calling Hailuo 2.3 as an API without building the integration
Direct API access to a video model means credentials, a submit-and-poll job pattern, retries, moderation rejections, and somewhere to store the output, and at over a minute per clip the polling part is not optional.
On a Treza AI video pipeline, Hailuo 2.3 is a node on a canvas. Set the duration to 6 or 10, wire in a first frame if the shot starts on a still, place it next to a script, voiceover, or captioning step, and publish the whole chain as one versioned endpoint. Your application calls a typed /invoke route with JSON in and JSON out, or points an existing OpenAI SDK at the OpenAI-compatible route.
The swap is what pays off later. When the next model wins on price or quality, and for this model that may already be Hailuo 3 Max, you change the dropdown and republish, and the endpoint your product calls does not change. Our hosted MCP server exposes the same pipelines as tools, so an agent in Claude or ChatGPT can render a brief and return the link. Unlike Hailuo 3 Max, which Treza also reaches through fal, Hailuo 2.3 runs on the OpenRouter lane only.
Where Hailuo 2.3 earns its place in a graph
Three jobs where a silent widescreen clip at a fixed price is enough. Establishing shots under narration for a faceless channel, where a 10 second wide shot at 82 cents opens a segment and the voice carries it. B-roll beats in an explainer, 6 seconds at 49 cents each, stitched by the Sequence node, which handles up to 12 shots. And matching an existing Hailuo 2.3 library.
The fastest way to learn what it costs for your own work is to run your real prompt and read the charge. Open the AI video generator, pick Hailuo 2.3, and generate once.
Frequently Asked Questions
How much does Hailuo 2.3 cost per second?
$0.0817 per second at every setting, the listed rate for the endpoint we run as of October 2026. A 6 second clip is about 49 cents of provider cost and a 10 second clip about 82, and our 6 second test billed exactly $0.4902. On Treza the charge comes from a prepaid credit balance, and a failed generation charges nothing.
What resolution does Hailuo 2.3 render?
The catalog lists 1080p and exposes no setting, but our 6 second test came back at 1366 by 768 at 24 frames per second. Plan on 768p-class output and upscale in the graph for a clip shown large.
How long can a Hailuo 2.3 clip be?
6 or 10 seconds, nothing in between and nothing longer. For longer pieces, stitch generations with the Sequence node, which handles up to 12 shots, or assemble them in the timeline editor.
Does Hailuo 2.3 generate audio?
No. Our test file carried no audio track, even on a prompt that asked for ambient sound. Plan sound in the graph with Text to Speech or Audio Generation, mixed in the Sequence node, or pick Hailuo 3 if the soundtrack has to come from the model.
How is Hailuo 2.3 different from Hailuo 3 Max?
Hailuo 3 Max is the newer MiniMax model: 5 to 15 second clips in six aspect ratios, a first and a last frame, 480p or 768p, and far faster renders in our tests. Hailuo 2.3 is 16:9 only, 6 or 10 seconds, first frame only, at one flat price. Pick 2.3 to match a library already rendered on it; otherwise start with 3 Max.
Can I call Hailuo 2.3 through an API without a MiniMax account?
Yes. On a Treza pipeline, Hailuo 2.3 is a node you configure once and publish as a versioned HTTP endpoint, with a typed /invoke route and an OpenAI-compatible route. A hosted MCP server exposes the same pipelines as tools for agents. Generations are metered from prepaid credits that start at $5 and never expire.
