> ## Documentation Index
> Fetch the complete documentation index at: https://docs.getbifrost.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# v1.5.0-prerelease7

> v1.5.0-prerelease7 changelog - 2026-04-29

<Tabs>
  <Tab title="NPX">
    ```bash theme={null}
    npx -y @maximhq/bifrost --transport-version v1.5.0-prerelease7
    ```
  </Tab>

  <Tab title="Docker">
    ```bash theme={null}
    docker pull maximhq/bifrost:v1.5.0-prerelease7
    docker run -p 8080:8080 maximhq/bifrost:v1.5.0-prerelease7
    ```
  </Tab>
</Tabs>

<Update label="Bifrost(HTTP)" description="1.5.0-prerelease7">
  ## ✨ Features

  * **Passthrough Streaming Accumulation** - Added accumulator for passthrough streaming responses, enabling proper logging and cost tracking on raw provider streams
  * **Auto-Resolve Provider** - Inference and integration routes now auto-resolve the provider when no provider prefix is given on the model name
  * **Per-Request Content Logging Overrides** - Opt-in per-request overrides for content logging and raw request/response visibility, with DB migrations and live-reload
  * **Unified Dimension Headers (`x-bf-dim-*`)** - New unified dimension headers automatically forwarded to logs, traces, Prometheus, and Maxim tags
  * **OpenAI Realtime Audio (Base64)** - Audio base64 encoding support for OpenAI realtime provider (thanks [@Mahmoud-Khater](https://github.com/Mahmoud-Khater)!)
  * **Local Cache Hit Rate Speedometer** - Dashboard speedometer showing local cache hit rate (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * **VK-Scoped Model Lists** - Model list endpoints now scoped to virtual-key-allowed providers and models via request headers
  * **MCP Reverse Proxy OAuth** - External base URL support for reverse-proxy MCP OAuth flows
  * **`schemas.Duration` Type** - Go duration string support for MCP, Redis, Weaviate, and mocker duration fields
  * **Finish Reasons in OTEL Root Spans** - Finish reasons added to root spans, with correct model and provider names propagated
  * **Routing Rules Scope Cache** - Cache routing rules per scope upfront, plus model-catalog routing engine label and icon

  ## 🐞 Fixed

  * **OTEL Cost Info** - Fixed cost info in OTEL calls and response tools
  * **Migrations Conflict Resolution** - Fixed migrations for conflicts
  * **WebSocket /responses Reliability** - WebSocket responses now working with improved logging, cost tracking, and VK stripping
  * **MarshalJSON Auto-Redaction** - Removed `MarshalJSON` auto-redaction; explicit redaction now applied to env-backed fields in `ProxyConfig`, `ClientConfig`, and `AzureKeyConfig`
  * **Vertex `google/` Prefix** - Strip `google/` prefix from Vertex model IDs across all request types
  * **Vertex Multi-Region Routing** - Multi-region-only models now route to multi-region endpoints when the provider key is configured for a single region only
  * **OAuth Token `expires_at`** - `expires_at` is now nullable; refresh/reconnect guarded on nil expiry
  * **OpenAI Responses Tool Fields** - Preserved tool fields in OpenAI responses (thanks [@princepal9120](https://github.com/princepal9120)!)
  * **Semantic Cache Determinism** - Deterministic request hashing and `CacheDebug` propagation in streaming (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * **Streaming Pool-Reuse Corruption** - Snapshot `RequestType` before closure to prevent pool-reuse corruption in streaming requests
  * **Self-Looping Chain Rules** - Chain rules with self-loops now continue evaluating subsequent rules instead of halting
  * **Default Routing Provider Filter** - Filter out unconfigured providers in default routing
  * **Network Config Fallback for Ollama/SGL** - Fall back to network config if key config URL is not set for Ollama and SGL
  * **`base_url` Backward Compatibility** - `base_url` added to `network_config` for backward compatibility
  * **Streaming Pipeline `RawRequest`** - Propagate `RawRequest` through streaming pipeline and fix pool leak (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * **Logging Streaming Errors** - Improved streaming error handling in logging plugin (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * **`governance_budgets` Join** - Corrected join condition to use `virtual_key_id`
  * **OTEL Input/Output Messages** - Fixed input/output messages propagation to root span
  * **`resolvePeriod` UTC** - Fixed UTC handling in `resolvePeriod` time calculation
  * **Dockerfile.local** - `Dockerfile.local` now uses local packages (thanks [@ReStranger](https://github.com/ReStranger)!)
  * **Semanticcache Provider Keys** - Inherit provider keys from global client in semanticcache plugin

  ## 🔧 Maintenance

  * **Helm Chart Upgrades** - Guardrails Helm chart upgrade; Helm `apply` step added; Kubernetes pod-discovery RBAC templates added
  * **Dashboard UI Polish** - Popover scrolling, sheets/cluster page indentation, save-button validation, dialog overflow, fixed `ChartCard` heights, broader `ComboboxSelect` adoption (pricing, routing, assignment fields)
  * **Plugin Lifecycle Logging** - Added log level param to `AppendRoutingEngineLog`; trimmed unused dependencies in semanticcache
  * **OpenAPI Regeneration** - Regenerated `openapi.json`
</Update>

<Update label="Core" description="1.5.6">
  * feat: add passthrough streaming accumulation
  * feat: add `x-bf-dim-*` unified dimension headers forwarded to logs, traces, Prometheus, and Maxim tags
  * feat: add `schemas.Duration` type with Go duration string support for MCP, Redis, Weaviate, and mocker duration fields
  * feat: add audio base64 encoding for OpenAI realtime provider (thanks [@Mahmoud-Khater](https://github.com/Mahmoud-Khater)!)
  * feat: auto-resolve provider when no provider prefix is given on inference and integration routes
  * feat: add opt-in per-request overrides for content logging and raw request/response visibility
  * feat: add finish reasons to root spans with correct model and provider names in OTEL
  * fix: fix cost info in OTEL calls and response tools
  * fix: remove `MarshalJSON` auto-redaction; apply explicit redaction for env-backed fields in `ProxyConfig`, `ClientConfig`, and `AzureKeyConfig`
  * fix: strip `google/` prefix from Vertex model IDs across all request types
  * fix: snapshot `RequestType` before closure to prevent pool-reuse corruption in streaming
  * fix: fallback to network config if key config url is not set for ollama and sgl
  * fix: route vertex multi-region-only models to multi-region endpoints when key is single-region
  * fix: make OAuth token `expires_at` nullable and guard refresh/reconnect on nil expiry
  * fix: preserve responses tool fields in openai (thanks [@princepal9120](https://github.com/princepal9120)!)
  * fix: deterministic semanticcache request hashing and `CacheDebug` propagation in streaming (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * chore: env var updates and bedrock cli compatibility changes
  * chore: add ollama and vllm key configs in llmtests
  * test: add roundtrip test for `input_audio_buffer.append` audio serialization
</Update>

<Update label="Framework" description="1.3.6">
  * feat: add passthrough streaming accumulation
  * feat: dashboard local cache hit rate speedometer (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * feat: add db migrations and live-reload for per-request content storage and raw override flags
  * feat: add external base url support for reverse proxy mcp oauth
  * feat: add opt-in per-request overrides for content logging and raw request/response visibility
  * feat: store and show finish reason in LLM calls
  * feat: add finish reasons to root spans with correct model and provider names in OTEL
  * feat: add `schemas.Duration` type with Go duration string support
  * fix: fix migrations for conflicts
  * fix: fix cost info in OTEL calls and response tools
  * fix: native WebSocket `/responses` working with improved logging and cost tracking
  * fix: remove `MarshalJSON` auto-redaction; explicit redaction for env-backed fields
  * fix: correct `governance_budgets` join condition to use `virtual_key_id`
  * fix: input/output messages propagation to root span for OTEL
  * fix: route vertex multi-region-only models to multi-region endpoints when key is single-region
  * fix: make OAuth token `expires_at` nullable and guard refresh/reconnect
  * fix: propagate `RawRequest` through streaming pipeline and fix pool leak (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * fix: deterministic semanticcache request hashing and `CacheDebug` propagation (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * chore: upgraded core to v1.5.6
</Update>

<Update label="compat" description="0.1.5">
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>

<Update label="governance" description="1.5.6">
  * perf: cache routing rules per scope upfront and add model-catalog routing engine label and icon
  * fix: allow self-looping chain rules to continue evaluating subsequent rules instead of halting
  * chore: guardrails helm chart upgrade
  * chore: bedrock cli compatibility changes
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>

<Update label="jsonparser" description="1.5.6">
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>

<Update label="logging" description="1.5.6">
  * feat: add passthrough streaming accumulation
  * feat: opt-in per-request overrides for content logging and raw request/response visibility
  * feat: store and show finish reason in LLM calls
  * feat: forward `x-bf-dim-*` unified dimension headers to logs and traces
  * fix: improve streaming error handling (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * fix: deterministic semanticcache request hashing and `CacheDebug` propagation in streaming (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>

<Update label="maxim" description="1.6.6">
  * feat: forward `x-bf-dim-*` unified dimension headers as Maxim tags
  * fix: deterministic request hashing for cache (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>

<Update label="mocker" description="1.5.6">
  * feat: add Go duration string support for mocker duration fields via `schemas.Duration`
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>

<Update label="otel" description="1.2.6">
  * feat: add finish reasons to root spans with correct model and provider names
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>

<Update label="prompts" description="1.0.6">
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>

<Update label="semanticcache" description="1.5.6">
  * feat: add Go duration string support for Redis and Weaviate fields via `schemas.Duration`
  * refactor: inherit provider keys from global client
  * fix: deterministic request hashing and `CacheDebug` propagation in streaming (thanks [@loss-and-quick](https://github.com/loss-and-quick)!)
  * chore: add log level param to `AppendRoutingEngineLog` and trim unused dependencies
  * chore: env var updates
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>

<Update label="telemetry" description="1.5.6">
  * feat: forward `x-bf-dim-*` unified dimension headers to Prometheus metrics
  * chore: upgraded core to v1.5.6 and framework to v1.3.6
</Update>