v2.4.1
Stable Release
Released: Oct 24, 2025

The Nebula Update

A major infrastructure and content expansion focusing on semantic graph connectivity, real-time translation accuracy, and the introduction of Aevum-Core v4.2.

Total Articles
2,418,502
↑ +12,400 since v2.4.0
Languages
143
↑ +3 new locales
AI Model
Core v4.2
Optimized Inference
Graph Nodes
890M
↑ 18% connectivity
🚀 New Features
  • New

    Semantic Graph v3 Visualization

    Enhanced 3D visualization engine for knowledge graphs. Users can now zoom into dense clusters, filter by confidence score, and export sub-graphs as PNG or SVG.

  • New

    Real-Time Collaborative Annotation

    Verified contributors can now annotate articles simultaneously. Changes are streamed via WebSocket and reconciled automatically using CRDTs.

  • New

    Neural Translation for 3 New Languages

    Added support for Yoruba, Quechua, and Georgian using the new Aevum-NMT engine. Includes back-translation verification.

⚡ Improvements
  • Improve

    Search Latency Reduction

    Re-indexed the primary search vector store. P99 latency dropped from 140ms to 85ms globally. Relevance scoring updated for technical queries.

  • Improve

    Mobile Reader Performance

    Implemented virtualized scrolling for long articles. Reduced bundle size by 40% on iOS and Android webviews.

🐛 Bug Fixes
  • Fix

    Citation Linking Errors in Mathematics

    Resolved an issue where LaTeX-heavy articles generated broken citation anchors. Fixed parser logic for nested equations.

  • Fix

    API Rate Limit Bypass

    Patched a race condition in the rate limiter that allowed burst requests to exceed quotas under high concurrency. Added stricter enforcement.

⚠️ Deprecation Notice

The /v1/graphql endpoint is deprecated and will be removed in v3.0.0. Please migrate to /v2/graphql by Dec 31, 2025. Schema changes are backward compatible.

Schema Updates

New fields added to the Article type for enhanced metadata retrieval.

type Article {
  id: ID!              # Global unique identifier
  title: String!       # Article title
  content: Content!    # Rich text body
  
  # NEW in v2.4.1
  graphRank: Float!   # Centrality score in knowledge graph
  confidence: Float!  # AI verification confidence (0.0-1.0)
  locales: [Locale!]! # Available language variants
  related: [Article!]! # Top 10 semantic neighbors
}

Rate Limits

d style="padding: 0.75rem; font-family: var(--mono);">60
Tier Requests / min Burst
Free 100
Pro 1,000 2,000
Enterprise Custom Custom
View Full API Documentation →

Download full dataset dumps, static exports, or model weights for this version. All files are signed and verifiable.

📦
34.2 GB
Full JSON Dump
Complete article database in structured JSON format. Includes metadata, content, and revision history.
Download Dump
📄
12.8 GB
Static HTML Export
Self-contained static website archive. Ready to deploy on any web server or offline reader.
Export HTML
🔗
8.4 GB
Knowledge Graph RDF
Semantic relationships and entities in Turtle/RDF format. Compatible with standard triplestores.
Download RDF
🧠
2.1 GB
Aevum-Core v4.2 Weights
Open-weight embedding model fine-tuned on encyclopedia corpus. PyTorch format.
Get Weights
SHA-256: 8f3a9c2d1e4b5f6a7c8d9e0f1a2b3c4d5e6f7a8b9c0d1e2f3a4b5c6d7e8f9a0b
GPG Key: Available at keys.openpgp.org?search=0xAevumRoot
⛔ Breaking Change

The search() function now returns results sorted by relevance_score instead of created_date. Default sort parameter removed. Explicit sort must be provided for legacy behavior.

Migration Guide: v2.4.0 → v2.4.1

Most applications will work without changes. However, if you rely on date-based sorting or specific GraphQL fields, review the following.

# BEFORE (v2.4.0)
curl -X POST https://api.aevum.com/v1/search \\
  -H "Authorization: Bearer $TOKEN" \\
  -d '{"query": "quantum", "sort": "date"}'

# AFTER (v2.4.1)
curl -X POST https://api.aevum.com/v1/search \\
  -H "Authorization: Bearer $TOKEN" \\
  -d '{"query": "quantum", "sort": {"field": "date", "order": "desc"}}'
View Full Migration Guide →