> ## Documentation Index
> Fetch the complete documentation index at: https://apixo.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Changelog: New Models, API Changes & Pricing Updates

> New models, API changes, pricing updates, and documentation improvements on APIXO

Follow the latest model launches, API updates, and documentation improvements on APIXO.

<Update label="2026-09" tags={["New models", "API updates", "Pricing", "Docs"]}>
  ## September 2026

  **New models**

  * [GPT Image 2.5 Flare](/docs/models/image/gpt-image-2-5-flare) and [GPT Image 2.5 Sunburst](/docs/models/image/gpt-image-2-5-sunburst) — text-to-image and reference-guided image editing with five quality tiers (2026-09-09)
  * [Z-Image LoRA Pro](/docs/models/image/z-image-lora-pro) — vivid-detail text-to-image generation with flexible `size`, common `aspect_ratio` presets, optional seed control, and prompt extension (2026-09-03)

  **API updates**

  * [Seedance 2.5](/docs/models/video/seedance-2-5) now supports a 480p Draft-to-1080p Final workflow. Generate review candidates with `draft=true`, then submit a selected Draft task ID as `draft_task_id`. For dynamic-duration Drafts, Final billing uses the Draft's actual output duration rounded to 3 decimal places. In a four-candidate workflow, generating four Drafts and one selected Final can save about 57% and roughly double iteration efficiency compared with four direct 1080p generations. (2026-09-23)

  **Pricing updates**

  * [Seedance 2.0 Mini](/docs/models/video/seedance-2-0-mini) public per-second pricing was restored to `$0.0533` / `$0.0280` for 480p without/with video reference and `$0.1067` / `$0.0607` for 720p without/with video reference. (2026-09-29)
  * [GPT Image 2.5 Flare](/docs/models/image/gpt-image-2-5-flare) and [GPT Image 2.5 Sunburst](/docs/models/image/gpt-image-2-5-sunburst) are billed per successful image from `$0.0100` to `$1.0000`, depending on `quality` and `resolution`.
  * [Z-Image LoRA Pro](/docs/models/image/z-image-lora-pro) is billed per successful generated image at `$0.025 / image`. Prompt extension does not add an extra public charge.
</Update>

<Update label="2026-08" tags={["New models", "Pricing", "Docs"]}>
  ## August 2026

  **New models**

  * [Seedance 2.5](/docs/models/video/seedance-2-5) — multimodal video generation with `text-to-video`, `first_and_last_frames`, `omni_reference`, dynamic duration, and `480p` / `720p` / `1080p` outputs
  * [Grok Imagine Video 1.5](/docs/models/video/grok-imagine-video-1-5) — image-to-video generation with 1-15 second outputs and `480p`, `720p`, or `1080p` pricing tiers
  * [MiniMax H3](/docs/models/video/minimax-h3) — 768p and 2K video generation for text-to-video, image-to-video, and multimodal reference-to-video workflows
  * [Wan 3.0 Video](/docs/models/video/wan-3-0-video) — text-to-video, image-to-video, and multimodal reference-to-video workflows with `480p`, `720p`, and `1080p` outputs
  * [Wan 3.0 Video LoRA](/docs/models/video/wan-3-0-video-lora) and [Wan 3.0 Video Pro LoRA](/docs/models/video/wan-3-0-video-pro-lora) — LoRA video generation with text-to-video, image-to-video, and reference-to-video workflows
  * [Kling 3.0 Omni](/docs/models/video/kling-3-0-omni) — standard, pro, and 4K text, image, and reference video generation workflows

  **Pricing updates**

  * [Seedance 2.5](/docs/models/video/seedance-2-5) is billed per second by resolution and video-reference usage: `480p` at `$0.085-$0.140 / second`, `720p` at `$0.190-$0.315 / second`, and `1080p` at `$0.500-$0.800 / second`. `duration=-1` is pre-charged as 30 output seconds and settled by actual output duration after success.
  * [Grok Imagine Video 1.5](/docs/models/video/grok-imagine-video-1-5) is billed per second: `480p` at `$0.02 / second`, `720p` at `$0.035 / second`, and `1080p` at `$0.048 / second`.
  * [Grok Video](/docs/models/video/grok-video) now supports `1080p` and uses updated per-second pricing: `480p` at `$0.02 / second`, `720p` at `$0.035 / second`, and `1080p` at `$0.048 / second`.
  * [Grok Image](/docs/models/image/grok-image) pricing docs now reflect the current `$0.10 / image` rate for both supported modes.
  * [Seedream 5.0 Pro](/docs/models/image/seedream-5-0-pro) now documents `layer-decomposition`, including text-based `<bbox>` region guidance in `prompt`, and the current `$0.045 / output image`, `$0.09 / output image`, and `$0.36` default layer-decomposition quote.
  * MiniMax H3 now supports `768p` at `$0.09 / second`; `2k` remains `$0.13 / second`. Reference video seconds use the selected output resolution rate, and reference images after the first 5 remain `$0.04 / image`.
  * [Wan 3.0 Video](/docs/models/video/wan-3-0-video) uses updated per-second pricing: `480p` at `$0.07`, `720p` at `$0.12`, and `1080p` at `$0.225` across supported modes.
  * [Wan 3.0 Video LoRA](/docs/models/video/wan-3-0-video-lora) is billed per second at `$0.0525`, `$0.1050`, and `$0.2100` for `480p`, `720p`, and `1080p`; [Wan 3.0 Video Pro LoRA](/docs/models/video/wan-3-0-video-pro-lora) is billed per second at `$0.1890`, `$0.2380`, and `$0.3220` for `1080p`, `2k`, and `4k`.
  * [Kling 3.0 Omni](/docs/models/video/kling-3-0-omni) is billed per second from `$0.084` for standard generation without sound up to `$0.42` for 4K generation.

  **Documentation**

  * [Seedance 2.5](/docs/models/video/seedance-2-5) now documents the final `data.billing` settlement snapshot returned for successful `duration=-1` tasks, including seconds units, 3-decimal rounding, input/output video counts, and repeated polling stability.
  * [Grok Image](/docs/models/image/grok-image) now documents the expanded `text-to-image` aspect ratios and the `image-to-image` limit of up to 5 reference images.
  * [Grok Video](/docs/models/video/grok-video) now documents direct `image-to-video` support for up to 7 reference images at `480p` / `720p`, with `1080p` limited to 1 reference image.
</Update>

<Update label="2026-07" tags={["New models", "Pricing", "Docs"]}>
  ## July 2026

  **New models**

  * [Seedream 5.0 Pro](/docs/models/image/seedream-5-0-pro) — flagship Seedream image generation with high-resolution output (2026-07-09)
  * [Z-Image LoRA](/docs/models/image/z-image-lora) — LoRA-based stylized image generation (2026-07-21)
  * [Wan 2.2 Video LoRA and Wan 2.7 Video LoRA](/docs/models/video/wan-2-7-video-lora) — LoRA fine-tuned video generation (2026-07-15)
  * [Runway Gen4 Image and Runway Gen4 Image Turbo](/docs/models/image/runway-gen4-image) — reference-guided image generation (2026-07-02)
  * Utility models: [Image Upscaler](/docs/models/image/image-upscaler), [Video Upscaler](/docs/models/video/video-upscaler), watermark removers, and swap tools, with updated pricing (2026-07-10)

  **Documentation**

  * New model category landing pages for image, video, and audio (2026-07-24)
  * Single-tree sidebar navigation with collapsible model families (2026-07-27)
</Update>

<Update label="2026-06" tags={["New models", "Pricing"]}>
  ## June 2026

  **New models**

  * [Seedance 2.0 Mini](/docs/models/video/seedance-2-0-mini) — lightweight Seedance video generation (2026-06-23)
  * [Kling 3.0 Turbo](/docs/models/video/kling-3-0-turbo) — faster Kling 3.0 video generation (2026-06-18)
  * [MiniMax Speech 2.8](/docs/models/audio/minimax-speech-2-8) — speech generation with preset voices (2026-06-16)
  * [Wan 2.7 Video](/docs/models/video/wan-2-7-video) — latest Wan video generation (2026-06-16)
  * [MiniMax Image 01](/docs/models/image/minimax-image-01) — MiniMax image generation (2026-06-12)
  * [Seedance 1.5 Pro Extend](/docs/models/video/seedance-1-5-pro) — extended duration support (2026-06-18)

  **Updates**

  * Seedance 2.0 adds 4K output support (2026-06-24)
  * Veo 3.1 modes and pricing updates (2026-06-09, 2026-06-16)
</Update>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.