# Llm Observability

Ranked by the Vioscale composite score: a confidence-weighted blend of independent signals, not user reviews.

| # | Software | Score | Confidence | Vendor |
|--:|---|--:|---|---|
| 1 | [Portkey](https://www.vioscale.ai/software/portkey) | 78.4 | 76% (high) | - |
| 2 | [Traceloop](https://www.vioscale.ai/software/traceloop) | 69.3 | 54% (medium) | - |
| 3 | [Lunary](https://www.vioscale.ai/software/lunary) | 69 | 75% (high) | - |
| 4 | [Langfuse](https://www.vioscale.ai/software/langfuse) | 63.8 | 55% (medium) | - |
| 5 | [Braintrust](https://www.vioscale.ai/software/braintrust) | 60.9 | 33% (low) | - |
| 6 | [LangSmith](https://www.vioscale.ai/software/langsmith) | 56.9 | 14% (low) | - |
| 7 | [Humanloop](https://www.vioscale.ai/software/humanloop) | 55 | 55% (medium) | - |
| 8 | [Weights & Biases Weave](https://www.vioscale.ai/software/weights-biases-weave) | 52 | 9% (low) | - |
| 9 | [AgentOps](https://www.vioscale.ai/software/agentops) | 48.8 | 55% (medium) | - |
| 10 | [Arize Phoenix](https://www.vioscale.ai/software/arize-phoenix) | 48 | 18% (low) | - |
| 11 | [Opik](https://www.vioscale.ai/software/opik) | 46.9 | 36% (low) | - |
| 12 | [Helicone](https://www.vioscale.ai/software/helicone) | 26.2 | 18% (low) | - |

---
*Source: Vioscale (https://www.vioscale.ai/categories/llm-observability).*