Changelog
Every notable change shipped to vMira. New entries appear at the top; older releases live below. Subscribe to the RSS feed to follow along.
- ·v1.2.1Improvement
SEO overhaul — hreflang, JSON-LD split, per-page schema, comparison and use-case landings
Fixed missing hreflang tags on all locale pages, split global JSON-LD so FAQ/WebApplication only emit on the home page, added Article/BreadcrumbList schema per help article, shipped /vs/[slug] and /use-cases/[slug] landing systems, switched several 307 redirects to 301.
- ·v1.2.0Feature
Audio/video voice mode — slice 1 backend landed
First voice/video AV chat backend integration shipped to production. Web frontend slice is in progress.
- ·v1.1.3Infrastructure
Migrated production to OVH BHS5
vMira production now runs on OVH dedicated infrastructure (192.99.43.112). Lower latency for North American users, full control over the data plane.
- ·v1.1.2Feature
Cross-chat shared context (per-profile memory)
Mira can recall what you discussed in previous chats when relevant. Cosine-similarity retrieval with time-decay, injected into the ai_proxy extra block. Opt-out per profile.
- ·v1.1.1Fix
Hotfix — model_chat join column
Resolved an issue where chat history occasionally failed to load due to a m.user_id vs c.user_id join error.
- ·v1.1.0Feature
Research Mode public beta
STORM + IterResearch hybrid for generating 60+ page research reports with sourced citations. Available on Plus and Pro plans.
- ·v1.0.5Feature
Finance subsystem — EDGAR, RSS, Wikipedia
Built-in finance information sources. Every response includes a no-advice disclaimer. No paid market-data vendor — public sources only.
- ·v1.0.4Feature
Container/context system
MemGPT-style expandable container engine added — the foundation for cross-chat shared context.
- ·v1.0.0Feature
vMira 1.0 — public launch
First general-availability release. Chat, code, design, music, document analysis, voice input, OpenAI-compatible API, 50+ interface languages, crypto payments.