Release Notes

Stay up to date with the latest features, improvements, and fixes across MegaNova services.

---

April 2026

Inference Platform

Released: April 2026

New Features

• Claude Opus 4.7 added to the catalog with native adaptive thinking support ($4.0/$20.0 per 1M tokens, 1M-token context, multimodal)

• Limits page in the console now separates platform-wide rate limits (RPM/TPM) from per-model daily limits (RPD) for easier scanning

Bug Fixes

• Gemma models now display the Google logo instead of the Gemini icon

---

March 2026

Inference Platform

Released: March 2026

New Models

• MiMo-V2-Pro (1T params, $1.0/$3.0 per 1M tokens) and MiMo-V2-Omni (multimodal, $0.4/$2.0) from Xiaomi

• MiMo-V2-Flash with optional reasoning toggle ($0.1/$0.3 per 1M tokens)

• Mistral-Small-4-119B multimodal with reasoning support ($0.15/$0.6 per 1M tokens)

New Features

• Direct thinking-param support for GLM-5

• Model detail page now shows the canonical model ID under the alias with a copy button

Bug Fixes

• Qwen3.5-Plus pricing corrected to use live API values; tiered context pricing restored on Gemini Pro models only

• Removed expired promotions for GLM-5 and DeepSeek V3.2

---

February 2026

Inference Platform

Released: February 2026

New Features

• Native MCP server launched at https://api.meganova.ai/mcp/ — Model Context Protocol with tools (chat/text completions, image/video/audio generation, embeddings, model discovery), resources (meganova://models/*, meganova://gpu/instances), and prompts (character_persona, image_generation_prompt, video_generation_prompt). See the MCP integration guide.

• POST /v1/images/edits — OpenAI-compatible image editing endpoint

• GET /v1/videos/generations/{id}/stream — SSE endpoint for streaming completed video output

• GET /v1/providers/health — public media-provider health snapshot

• Webhook events now fire on video generation state transitions; per-user webhook CRUD is active

New Models

• Gemini 3.1 Flash Image Preview (Nano-Banana-2) via Vertex AI

• BytePlus Seedream 5.0 Lite with support for up to 14 reference images

• Alibaba Wan 2.6 — I2V-Flash, R2V, and R2V-Flash variants

• Alibaba Z-Image-Turbo integration

Improvements

• GET /v1/models now requires API-key auth (anonymous access deprecated with sunset headers)

• OpenAI SDK compatibility improved across /v1/models, /chat/completions, and /embeddings

• Console: Claude and Gemini playground examples now surface thinking/reasoning-effort parameters; tier table shows RPM/TPM

Bug Fixes

• Mixing local and URL reference images in image-edit models now works

• Nano Banana provider options wrapped correctly to prevent 502s

• Seedance Lite I2V image handling fixed to use first_frame_image_uri

---

January 2026

Character Studio v1.5.0

Released: January 23, 2026

New Features

• Redesigned login page with split-screen layout and modern dark theme

• Added Google Sign-In button (coming soon)

• New Documentation link in sidebar for quick access

• Updated footer with comprehensive legal and resource links

Improvements

• Enhanced form validation with whitespace trimming

• Improved responsive design for mobile devices

• Better error handling for authentication flows

Bug Fixes

• Fixed footer links pointing to placeholder URLs

• Updated copyright year to 2026

---

December 2025

Character Studio v1.4.0

Released: December 2025

New Features

• Public landing page for unauthenticated users

• Workflow Editor with visual node-based character orchestration

• Creator Center with referral tracking and analytics

Improvements

• Enhanced character export with CCv3 specification

• Improved image generation with better CORS handling

• Updated pricing page with clearer tier information

---

November 2025

API Platform Updates

Released: November 2025

New Models

• Added Manta Pro series with enhanced character-dialogue capabilities

• New embedding models for improved semantic search

Improvements

• Reduced latency for serverless model inference

• Enhanced rate limiting with better error messages

---

How to Stay Updated

Subscribe to Updates

• Follow our Blog for detailed announcements

• Join our Discord community

Feedback

Have feedback or feature requests? Contact us at [email protected] or through Discord.

---

Version History

Release	Date	Highlights
Apr 2026	Apr 2026	Claude Opus 4.7, low-balance alerts, smarter RPD rollback
Mar 2026	Mar 2026	MiMo-V2 series, Mistral-Small-4, reseller management, vendor uptime
Feb 2026	Feb 2026	Native MCP server, image edits, webhook events, Seedream 5 Lite, Wan 2.6
v1.5.0	Jan 2026	Login redesign, footer updates (Character Studio)
v1.4.0	Dec 2025	Landing page, Workflow Editor (Character Studio)
v1.3.0	Nov 2025	Creator Center, referral system (Character Studio)
v1.2.0	Oct 2025	Character versioning, lorebook system (Character Studio)
v1.1.0	Sep 2025	Chat improvements, persona support (Character Studio)
v1.0.0	Aug 2025	Initial release (Character Studio)