Changelog

​

BYOK — Bring Your Own Midjourney Account

• Bind your own paid Midjourney subscription: fast-mode image tasks (imagine, variation, upscale, reroll, blend, describe) execute on your own account and consume zero platform credits. Everything else (video, edit tools) stays pay-as-you-go on the platform pool
• Flat credit rent: 20,000 credits/month per bound account (≈ $20), charged from your credit balance at binding, renewing monthly until you unbind. Existing bindings at launch received a 30-day courtesy period instead of a retroactive charge
• Multi-account pools: bind several accounts — each is billed separately and tasks route to the least-busy healthy account. Effective fast-image concurrency scales as max(6, min(account limit, 3 × usable accounts))
• No silent fallback: if your session expires or your fast hours run out, tasks fail with a clear error and actionable guidance — they never silently switch to paid platform usage. Unbind anytime to return to pay-per-use
• Dashboard-first setup: everything lives under My MJ Account — bind, live status with operator remedies, token refresh, per-account rent state, and unbind. Task Logs now mark BYOK vs platform tasks and chart daily consumption by source
• Same API: endpoints, parameters and your API key are unchanged; binding only changes where fast-mode image tasks execute
• Learn more: the BYOK product page and the BYOK guide cover prerequisites, failure states and recovery paths

​

Midjourney V8.2 Support

• V8.2 model: add —v 8.2 to your prompt. Midjourney released V8.2 on July 24, 2026 as an aesthetics update and published no parameter list or benchmarks, so everything below is behavior we probed against both upstream providers on July 27
• Parameter surface identical to V8.1: twelve flags tested. V8.2 accepts exactly what V8.1 accepts and rejects exactly what V8.1 rejects, so migrating is a one-flag change
• Native 2K via —hd: 2048×2048, billed at 1.5× (120 credits) vs 1× (80) for the 1024×1024 default. Verified by reading the PNG headers off delivered images, not by assumption
• Draft mode: —v 8.2 —draft returns a 24-image batch at the standard rate
• Rejected flags: —turbo, —oref / —ow, —cref / —cw, —q / —quality are rejected with an HTTP 400 naming the flag, before the job is submitted or billed
• Grid actions: like the rest of the V8 family, V8.2 grids support upscale, variation and reroll only — available_actions no longer advertises inpaint, outpaint, pan, remix or edit for these tasks
• Pricing unchanged: 80 credits per standard job, 120 with —hd, identical to V8.1
• API default is still v7: a prompt with no —v flag is unaffected by this release. We never inject a version on your behalf
• Same endpoints: no integration changes required

​

Midjourney V8.1 Support

• V8.1 Model: Add —v 8.1 to your prompt to use Midjourney’s latest alpha model — SD (1024) by default with ~40% faster generation than V8 HD
• Native 2K via —hd: add —hd for 2048×2048 output. See the current parameter matrix for the billing multiplier
• Image prompts and image weights restored: V8.1 supports image URLs in prompts and the —iw parameter for image weighting
• Prompt Shortener: Prompts that exceed length limits are automatically shortened by Midjourney
• Improved Describe: V8.1 produces longer, more detailed prompt descriptions
• Removed in V8.1: —q / —quality, —cref / —cw, and —oref / —ow are no longer supported. Requests containing these flags will be rejected with a clear error. Correction (Aug 10, 2026): this entry originally listed —no as removed — that was wrong. Midjourney’s compatibility chart and the current parameter matrix both show —no working on V8.1 and V8.2
• V8.0 deprecation: Midjourney plans to sunset V8.0 approximately 2 weeks after V8.1 stabilizes. Migrate —v 8 prompts to —v 8.1 to avoid disruption. Update (Aug 10, 2026): upstream retired V8.0 on July 24, 2026 — see Models for current guidance
• Same endpoints: V8.1 uses the existing /v1/diffusion, /v1/variation, /v1/reroll, /v1/remix, /v1/blend, and /v1/video-diffusion endpoints — no integration changes required

​

Midjourney V8 Support

• V8 Model: Add —v 8 to your prompt to use Midjourney’s latest model — ~5x faster generation, native 2K HD resolution, dramatically improved text rendering, and accurate anatomy
• Native 2K with —hd: V8 renders natively at 2K resolution. No upscale step needed. Use —hd for high-resolution output (4x credit cost)
• New —oref parameter: Omni Reference replaces —cref in V8, handling both person likeness and object form in a single parameter
• Enhanced coherence with —q 4: New quality mode for complex scenes with many elements (4x credit cost)
• V8 Blend: Blend endpoint now uses V8 natively for improved results
• Full follow-up support: Variation, reroll, remix, and animate all work on V8-generated images
• Backward compatible: Your existing V7 personalization profiles and style references work in V8

​

Topaz Integration Released

• Topaz Upscale API: /v1/enhance-upscale launched for ultra-high-resolution enhancement. This is a historical release note; the Enhancement API is currently suspended and returns HTTP 503.
• Improved Parameter Organization: Clear input method options (direct image URL or Midjourney task) with intuitive common parameters grouping
• Transparent Pricing: Enhanced resolution options table with cost breakdown and recommended use cases for each resolution tier

​

Python SDK Released

• Official Python SDK: Available on PyPI as legnext
• Complete Documentation: New SDKs tab with guides, API reference, and examples
• Features: Type hints, async support, webhook callbacks, error handling

​

New API endpoints

• /blend: Easily blend images together using Midjourney’s /blend command
• /describe: Writes four example prompts based on an image you upload
• /shorten: Submit a long prompt and receive suggestions on how to make it more concise

​

Response structure enhancements

• 4-grid image display: Enhanced response structure to support 4-grid image display for /imagine tasks, providing better visualization of generated image variations

​

Service optimization

• Temporary API endpoint suspension: The following endpoints are temporarily paused:
  • /upload-paint
  • /retexture
  • /remove-background
  • /enhance

  These endpoints will be restored promptly after optimization is complete.