Introducing LTX 2.3 Quality API and MAI Image 2.5 API on Pixazo API

Pixazo API is expanding its model lineup with two new high-quality additions: LTX 2.3 Quality API from Lightricks and MAI Image 2.5 API from Microsoft. The pairing brings premium text-to-image and AI video generation into the same developer-friendly stack, so builders can ship richer creative workflows without juggling separate vendor integrations.

This guide walks through what each model does, where its strengths sit, the trade-offs to plan for, and how teams typically combine them in production.

What Is LTX 2.3 Quality API?

LTX 2.3 Quality is the highest-fidelity variant of Lightricks' LTX-2 family of AI video generation models. Where LTX-2 introduced unified audio-visual generation in a single pass, the 2.3 Quality tier focuses on rendering cleanly at higher resolutions, holding motion coherence over longer clips, and reducing the subtle artifacts that often appear in fast-moving scenes or detailed textures.

Through Pixazo API, LTX 2.3 Quality is exposed alongside LTX 2.3, LTX 2 Pro, and LTX 2.0 — letting you pick the speed/quality balance that fits each project without switching providers.

What Is MAI Image 2.5 API?

MAI Image 2.5 is the latest iteration of Microsoft's MAI (Microsoft AI) Image text-to-image model line. The 2.5 release is built around scalable, production-grade image generation: it returns sharper edges, more reliable typography in posters and banners, and steadier results on long, structured prompts (style + subject + camera + lighting), which is where most consumer image models still drift.

On Pixazo API, MAI Image 2.5 is delivered through the same authentication, queue, and webhook flow as every other Pixazo model — so a team already calling Flux API, Seedream API, or Nano Banana API can route prompts to MAI Image with a single endpoint change.

Key Capabilities of LTX 2.3 Quality API

LTX 2.3 Quality API is tuned for situations where the cost of a slightly slower render is worth the quality bump. Specifically:

• Native high-resolution output with stable frame-to-frame consistency, reducing the flicker and texture-crawl common in lower-tier video models
• Smooth, physically plausible motion in scenes with multiple subjects, camera moves, or particle effects
• Unified audio + visual generation in one inference pass, so ambient sound and scene action stay in sync
• Extended-duration support for short-form storytelling, product demos, and B-roll without stitching shorter clips together
• Natural-language prompt understanding for camera direction (push-in, dolly, low-angle, rack focus, etc.)
• Image-to-video conditioning that respects the input image's composition rather than drifting away from it

Honest limitation: at LTX 2.3 Quality, render times are longer than LTX-2 Fast or LTX-2 Pro — plan for higher latency in high-throughput pipelines, or route bulk previews to a lighter tier first.

Key Capabilities of MAI Image 2.5 API

MAI Image 2.5 API is positioned for teams that need image generation to behave predictably across thousands of prompts — not just one beautiful hero image:

• Strong adherence to long, multi-clause prompts (subject, style, palette, framing, mood) without losing earlier instructions
• Notably improved text rendering for posters, banners, product mock-ups, and slide visuals
• High-resolution outputs suitable for marketing, e-commerce hero images, and editorial use
• Stable composition across batch generations of the same prompt — useful for A/B tests and ad sets
• Microsoft-backed safety filtering and content guardrails out of the box

Honest limitation: highly stylized "artist signature" looks (for example, hand-painted illustration with very specific brush logic) can still feel cleaner from specialized open-source models like Flux. Treat MAI Image 2.5 API as the "general-purpose, production-reliable" lane, not the "wild experimental" lane.

Suggested Read: Introducing Ideogram v4 API on Pixazo

What You Can Build With LTX 2.3 Quality API?

LTX 2.3 Quality API is best suited for short-form video work where production-grade fidelity matters more than raw render speed:

• Short-form social video — Reels, Shorts, and TikToks with consistent visual identity and clean motion
• Programmatic ad creative — animate product hero shots for paid social and display campaigns
• Landing-page hero loops — 6–10 second motion segments that loop seamlessly behind site copy
• E-commerce product video — turn flat catalog stills into short, on-brand motion clips
• Editorial motion teasers — accompany articles or newsletters with a coherent short scene
• Storyboard animations — bring concept frames to life to communicate scene intent in pitches

Suggested Read: Introducing LTX-2 Video API on Pixazo

What You Can Build With MAI Image 2.5 API?

MAI Image 2.5 API shines wherever predictable, scalable text-to-image generation is needed across many prompts and many runs:

• Marketing hero images — campaign visuals with reliable typography for banners, posters, and ads
• E-commerce product imagery — clean, brand-consistent product mock-ups at scale
• Social media graphics — quote cards, announcement posts, event banners with embedded text
• Editorial illustration — article hero images and section visuals with controllable style
• Ad creative variations — A/B test sets where composition stays stable across prompt tweaks
• Slide and deck visuals — internal and external presentations with clean, on-message imagery

Suggested Read: Introducing Nano Banana 2 API on Pixazo

Accessing LTX 2.3 Quality API on Pixazo API

LTX 2.3 Quality is available through the standard Pixazo API documentation: authentication via API key, request/response schemas, queue handling, webhooks for long-running renders, and consistent error codes alongside the rest of the LTX-2 family (LTX 2.3, LTX 2 Pro, LTX 2.0).

New accounts receive $5 in free credits on first payment — enough to run a meaningful evaluation against real prompts before committing to a workload.

Suggested Read: Introducing LTX 2 19B API on Pixazo

Accessing MAI Image 2.5 API on Pixazo API

MAI Image 2.5 is exposed through the same Pixazo API surface as every other supported model: a single API key, the same auth and webhook flow, identical error model, and per-request pricing visible directly on the model's documentation page.

Teams already calling other Pixazo image models can route prompts to MAI Image 2.5 with a single endpoint change — no separate Microsoft account required.

Suggested Read: Introducing ByteDance Seedream 4.5 API on Pixazo

Frequently Asked Questions about LTX 2.3 Quality API and MAI Image 2.5 API

1. What is LTX 2.3 Quality API?

LTX 2.3 Quality API is the highest-fidelity variant of Lightricks' LTX-2 family of AI video generation models, exposed through Pixazo API. It produces cleaner high-resolution output, smoother motion in complex scenes, and stronger frame-to-frame consistency than the lighter LTX-2 tiers, with unified audio + visual generation in a single inference pass.

2. What is MAI Image 2.5 API?

MAI Image 2.5 API is the latest iteration of Microsoft's MAI (Microsoft AI) Image text-to-image model, available through Pixazo API. It is built for scalable, production-grade image generation with sharper edges, more reliable typography for posters and banners, and steadier results on long, structured prompts.

3. Who developed LTX 2.3 Quality and MAI Image 2.5?

LTX 2.3 Quality is developed by Lightricks as part of its LTX-2 video model family. MAI Image 2.5 is developed by Microsoft as part of its MAI (Microsoft AI) image model line. Both are accessed through Pixazo API.

4. What is the difference between LTX 2.3 Quality and the earlier LTX-2 variants?

LTX 2.3 Quality is the highest-fidelity tier in the LTX-2 family — tuned for cleaner high-resolution output, smoother motion in complex scenes, and stronger frame-to-frame consistency. LTX-2 Pro and LTX-2 Fast remain better fits when latency matters more than top-end visual quality.

5. Does LTX 2.3 Quality generate audio with the video?

Yes. Like the rest of the LTX-2 family, LTX 2.3 Quality supports unified audio + visual generation in a single inference pass, so ambient sound and on-screen action stay in sync without a separate audio step.

6. Can LTX 2.3 Quality take an image as input?

Yes. The model supports both text-to-video and image-to-video generation, with image-to-video conditioning that respects the input image's composition rather than drifting away from it.

7. How does MAI Image 2.5 handle text rendering in images?

Text rendering is one of its strongest improvements over earlier MAI versions — letters stay readable in posters, banners, product mock-ups, and slide visuals where most consumer image models still struggle.

8. Can I use MAI Image 2.5 output commercially?

Yes. MAI Image 2.5 is intended for commercial use through Pixazo API's paid tiers, with Microsoft-backed safety filtering and content guardrails enabled by default. Review the model's licensing notes on its documentation page before shipping production content.

9. How are LTX 2.3 Quality API and MAI Image 2.5 API priced?

Each model has its own per-request pricing visible directly on its documentation page. Usage is billed against your Pixazo API balance, so a single payment method and API key cover both LTX 2.3 Quality and MAI Image 2.5 alongside the rest of the Pixazo model catalog.

10. How do I get started with these APIs?

Create a Pixazo API account, claim the $5 free credit on first payment, then call either model through the documented endpoints — no separate Lightricks or Microsoft account required. Both endpoints share the same authentication, queue, webhook, and error model as the rest of the Pixazo catalog.

Deepak Joshi - Content Marketing Specialist at Pixazo

Deepak Joshi is a Content Marketing specialist having a combined experience of 10+ years working in the digital world. He is one of the active contributors to Pixazo Blog.