I'm unhappy with separate logs across AI vendors; what should I use?

If you want one place to see and manage logs across multiple AI vendors, you should look for an LLM observability / gateway / unified logging layer rather than vendor-specific dashboards.

Good options by use case

1. Want centralized logs, tracing, costs, and prompt/version tracking

• LangSmith
• Helicone
• Arize Phoenix
• Braintrust
• Langfuse

These are built to collect requests/responses, latency, token usage, errors, evaluations, and traces across OpenAI, Anthropic, Gemini, etc.

2. Want to control traffic across vendors with logging built in

• LiteLLM Proxy
• Portkey
• OpenRouter (more routing than observability, but useful)

These act like a single API gateway for multiple providers, so your app talks to one endpoint and logs are unified.

3. Want enterprise-grade observability

• Datadog LLM Observability
• New Relic
• Splunk
• OpenTelemetry-based setup

Best if you already use enterprise monitoring and want AI telemetry alongside your existing infra.

My practical recommendation

• If you're a team building AI products: Langfuse or Helicone
• If you want a proxy that standardizes all vendor calls: LiteLLM Proxy
• If you want best-in-class experiment tracing/evals: LangSmith
• If you want enterprise monitoring integration: Datadog

What to look for

Make sure the tool supports:

• multiple model providers
• request/response capture
• token and cost tracking
• latency and error logs
• tracing across tools/workflows
• redaction / privacy controls
• evals / prompt versioning

If you want, I can give you a “best option for your stack” recommendation if you tell me:

1. which vendors you use,
2. whether you want logs only or routing too,
3. your language/framework (Python, JS, etc.).