Releasesby Glyphtruss 0.12.7
- add grpc-specific block for nginx server in front of custom images Release 0.12.7
Also shipped Jan 16, 2026Baseten in January 2026 →
🤖 New LLM Models Support
We've added support for new LLM models in our Voice Agent API! Available Models: OpenAI GPT 5.1 Chat (gpt-5.1-chat-latest) OpenAI GPT 5.1 (gpt-5.1) Anthropic Claude Sonnet 4.5 (claude-sonnet-4-5 Google Gemini 3 (gemini-3-pro-preview) Implementation: Configure your chosen model in your Voice Agent settings: { "type": "Settings", "agent": { "think": {...
Neon Auth on Vercel previews, neon init enhancements, and more release - Jan 16, 2026
Neon Auth on Vercel previews. Both the Vercel managed and Neon managed integrations now automatically provision Neon Auth on preview branches when enabled on yo...
What's new in Sigma
AI apps File upload columns, controls, and form fields (Beta) Use file columns in input tables to upload images, documents, and videos directly to Sigma. You can also use file upload controls to select files and upload them to destination input tables.
Enforce 2FA, CLI command structure, Singapore Buckets
Last week we wrapped our planning for Q1 2026, so we're officially back to our regularly scheduled programming. This week: a major security upgrade for Pro workspaces, a structural overhaul of the CLI, Object Storage now available in Singapore, smoother database workflows in the dashboard, and Editable Canvas Arrows graduating to GA. Let's get into it!
Preview mode support for MPA's
Preview mode support for MPA's (multi-page apps) Preview mode didn't really work for MPA's (multi-page apps) before. Now it does :) This should make initial activation, continuous iteration, and testing much easier for customers with Multi page apps Where: G&S detail page -> preview Available in EUDC: Yes When: This improvement is available now
Multiple LLM Provider Support
We've added new functionality that allows users to specify multiple LLM providers for your Voice Agent, ensuring your agent will automatically fallback to another provider should you experience any issues. The think object supports both a single provider and an array of providers. LLM providers will be used in the order that you specify them.
More from trussEvery truss release →
truss 0.12.6
- Support R2 weights sources Release 0.12.6
truss 0.12.5
- feat(weights): add https:// scheme support to WeightsSource fix: exclude weights from build hash to prevent unnecessary rebuilds Rename CLAUDE.md to AGENTS.md, symlink CLAUDE.md Support configurable run-as user ID for custom-server deployments Release 0.12.5
truss 0.12.4
- feat(truss): Move chain secret/api key to backend update deprecation message feat(watch): Add team disambiguation to truss watch command feat(teams): Indicate default team and sort teams alphanumerically in team selection feat(teams): Add team to remote config in trussrc Release 0.12.4
truss 0.12.8
- feat(chains): add MDN weights support to chains fix(chains): use chainlets_only param for chains watch model lookup Release 0.12.8
truss 0.12.9
- detect dup keys in yaml Mf/engine-builder-update Update contract for newest version of performance_client moe parallel Full SKU name instance type feat: LLM validation better validation Add a "truss upgrade" Update trt_llm_config.py Full SKU name for instance type for Chains Notify of upgrades once a day only Release 0.12.9
truss 0.12.10
- Remove Unnecessary query for model versions Release 0.12.10
truss 0.12.3
- Migrate truss watch GQL queries from model name to ID Weights config Schema feat(truss): Use client-side filtering for model lookup by name and team Support warnings in truss validation Release 0.12.3
truss 0.12.2
- feat(truss): support multiline docker_server.start_command Release 0.12.2