Skip to main content
npx -y @maximhq/bifrost --transport-version v1.4.15
​
Bifrost(HTTP)
1.4.15

​
✨ Features

  • Custom SSE Stream Reader — Replaced fasthttp’s default stream reader with a custom implementation to reduce bursts in SSE streaming
  • MCP Config Validation — Added validation for MCP tool configurations in config.json
  • Max Open Connections — Exposed max-open-connections for provider domains as a configurable field
  • Dashboard improvements - Added a bunch of new tabs and new graphs to the dashboard including Model Ranking, Cache usage, MCP usage etc
  • Improved dashboard and logs performance - We have improved LLM logs and Dashboard UI performance (~1400x faster) for large number of logs.

​
🐞 Fixed

  • Passthrough Streaming — Fixed passthrough streaming responses being buffered instead of streamed (thanks @jasonwells!)
  • MCP Notifications — Fixed MCP notifications returning incorrect status code (thanks @tcx4c70!)
  • Streaming Function Calls — Fixed function_call items not included in streaming response.completed output (thanks @majiayu000!)
  • Bedrock API Key Auth — Fixed Bedrock API key authentication without requiring bedrock_key_config (thanks @eliasbiondo!)
  • Bedrock Token Count Fallback — Added fallback to estimated token count when count-tokens API is unsupported (thanks @Edward-Upton!)
  • Anthropic Thinking Fixes — Fixed OpenAI-to-Anthropic-to-OpenAI thinking content conversion
  • Anthropic Header Selection — Fixed Anthropic header selection across providers
  • Gemini OpenAI Integration — Fixed Gemini flow for OpenAI-compatible integration
  • Semantic Cache Hashing — Fixed deterministic tools_hash and params_hash in semantic cache (thanks @ragokan!)
  • Anthropic Compaction — Added compaction support for Anthropic provider
​
Core
1.4.12
  • feat: added compaction support for Anthropic provider
  • feat: MCP config validation for configs in config.json
  • feat: exposed max-open-connections for provider domains as a configurable field
  • fix: include function_call items in streaming response.completed output (thanks @majiayu000!)
  • fix: Bedrock API key authentication without requiring bedrock_key_config (thanks @eliasbiondo!)
  • fix: Bedrock fallback to estimated token count when count-tokens API is unsupported (thanks @Edward-Upton!)
  • fix: OpenAI-to-Anthropic-to-OpenAI thinking content conversion
  • fix: Anthropic header selection across providers
  • fix: Gemini flow for OpenAI-compatible integration
  • fix: removed map marshals from core for improved serialization
  • fix: PG query optimizations
​
Framework
1.2.31
  • feat: added model catalog support
  • feat: MCP logs flow for high-volume log ingestion
  • fix: dashboard and logs list query optimizations
  • fix: PG query optimizations for materialized views
  • fix: truncate names exceeding 64 characters
  • chore: weaviate upgrade
​
governance
1.4.31
  • chore: upgraded core to v1.4.12 and framework to v1.2.31
​
jsonparser
1.4.30
  • chore: upgraded core to v1.4.12 and framework to v1.2.31
​
litellmcompat
0.0.20
  • chore: upgraded core to v1.4.12 and framework to v1.2.31
​
logging
1.4.31
  • chore: upgraded core to v1.4.12 and framework to v1.2.31
​
maxim
1.5.29
  • chore: upgraded core to v1.4.12 and framework to v1.2.31
​
mocker
1.4.30
  • chore: upgraded core to v1.4.12 and framework to v1.2.31
​
otel
1.1.30
  • chore: upgraded core to v1.4.12 and framework to v1.2.31
​
semanticcache
1.4.29
  • fix: deterministic tools_hash and params_hash for consistent cache lookups (thanks @ragokan!)
​
telemetry
1.4.31
  • chore: upgraded core to v1.4.12 and framework to v1.2.31