Quarter 4 2025: Major Platform Updates
Three months of updates: 15+ new models, team management, referral program, tiered pricing, data retention, new providers, and much more.
Here's everything new in LLM Gateway over the past three months — from cutting-edge models to powerful platform features.
🤖 New Models
👥 Team Management
Full team collaboration is now available:
- Roles: Owner, Admin, and Developer with granular permissions
- Seat-based pricing: 3 included seats, additional seats at $10/seat/month
- Prorated billing: Fair charges based on your billing cycle
- Enterprise options: SSO/SAML, SCIM provisioning, audit logs for larger teams
🏢 Admin Dashboard
New internal admin dashboard for platform management with:
- Token metrics calculation and aggregation
- User and organization management
- Performance monitoring
💰 Referral Program
Earn 1% of every referral's spending — forever!
- Automatic tracking via referral links and cookies
- Earnings credited to your account
- No cap on earnings
Find your referral link in Settings → Referrals.
🛡️ Routing Improvements
- Uptime-aware routing — Automatically routes away from low-uptime providers
- X-No-Fallback header — Prevent automatic provider fallback when needed
- Provider metrics — Routing decisions include provider performance data
- Multi-API key support — Load balance across multiple provider keys
- Latency optimization — Only consider latency when streaming is enabled
📊 Tiered Pricing
Volume-based pricing is now available:
- The more you use, the more you save
- Tiered cache pricing for supported models
- Discounts calculated automatically based on usage
- View pricing tiers on any model's detail page
🔒 Data Retention & Storage
New data retention controls with billing:
- Metadata-only (default): Only request metadata is stored
- Full retention: Store complete request and response data
- Data storage billing: Pay only for what you store
- Data limits: Set maximum storage limits per organization
Configure in Settings → Organization → Data Retention.
🔍 Model Discovery
Global Model Search
Search bar in the top navigation — instantly find any model across all providers.
Model Timeline & Metadata
- Release dates (
releasedAt,publishedAt) for all models - Model descriptions on detail pages
- Visual indicators for free models and provider mappings
- Filtered counts showing matching models and providers
⚖️ Model Comparison
Compare models side-by-side with our new comparison page:
- Input/output token pricing — Compare costs across providers
- Feature comparison — See capabilities like streaming, vision, tools at a glance
- Context windows — Compare model limits instantly
🎮 Playground Enhancements
- Comparison mode — Compare responses from multiple models side-by-side
- Group chat — Chat with multiple models simultaneously
- GitHub MCP tools — Integrated GitHub operations via Model Context Protocol
- Redirect after signin — Return to your original URL after authentication
- Image configuration — Set image parameters and reasoning effort options
📈 Analytics & Metrics
- Cost breakdown — Detailed breakdown in usage API responses
- Discount savings visualization — See savings on dashboard
- Throughput metrics — Monitor tokens per second
- Provider tracking — Track which providers serve requests
- Real-time metrics — Current minute history calculation
- Reasoning tokens — Included in cost calculations
💳 Billing Enhancements
- PDF invoices — Automatically generated and emailed for subscriptions and credits
- Company address on invoices
- Billing email management — Set separate billing contact
- Top-up credits button — Quick access to add credits
- Refund tracking — Full visibility in transaction history
- First-time bonus — Bonus credits on first purchase
🔐 Security Enhancements
- Email verification required — New accounts must verify before API access
- Enhanced email validation — Blocks plus signs, disposable emails, blacklisted domains
- Masked provider keys — Tokens masked in API responses
- Content filter detection — Proper error handling for filtered content
💸 Discounts
- 20% off all Google models — Applied automatically
- 10% off all Z.ai models — Applied automatically
- 75% off CanopyWave models — Qwen3-Coder, MiniMax M2, GLM-4.6, Kimi K2 Thinking
- 90% off DeepSeek V3.1 via CanopyWave partnership
🌐 New Provider Support
We've added support for major cloud providers and new partners:
- AWS Bedrock — Access Llama and other models via Amazon's managed service
- Microsoft Azure — Native Azure OpenAI integration
- Google Vertex AI — Full Vertex AI support with global routing
- CanopyWave — Discounted access to DeepSeek, Kimi, and more
- NanoGPT — New provider option
📱 API Enhancements
- Effort parameter — Control reasoning depth for Claude Opus 4.5
- no_reasoning flag — Disable reasoning when not needed
- Responses API — Support for GPT-5 Pro and GPT-5.1 Codex
- Thought signatures — Included in Google tool calls
- Custom upload limits — Configure per-gateway limits
- Unsupported message handling — Clear feedback for unsupported features
- JSON schema response format — Structured output with
json_schemaresponse format - JSON Output filter — Filter models by JSON output support in UI
- OpenAI JSON schema conversion — Automatic format conversion for Google models
- DeepSeek JSON schema — Native JSON schema output support for DeepSeek V3.1
📚 Documentation
- Cline integration guide — Use LLM Gateway with Cline
- Cursor guide — IDE integration documentation
- Data retention & caching docs — Full documentation
- Reasoning docs — Guidelines for reasoning models
- Collapsible links — Improved navigation
- GitHub timestamps — Last edit dates on all pages
📧 Email & Notifications
- Transactional emails — Trial start and cancellation notifications
- PDF invoice emails — Automatic invoice delivery
- Onboarding status — Synced with Brevo for marketing
🎉 Other Improvements
- Organization name editing — Update org name in preferences
- Free model rate limits — Tiered rate limiting for free models
- Sensitive word check — Parameter support for Z.ai
- Decimal.js — Precise cost calculations
- Shared UI package — Reusable components across apps
📢 Pro Plan Pricing Update
Starting December 19th, 2025, Pro plan fees will increase from 0% to 1%. This remains 80% less than the 5% fee on the Free plan, ensuring you continue to get exceptional value as you scale.