LangSmith vs DeepEval / Confident AI
Facts come from each vendor's own pages; download numbers are the headline package of each tool in 2026-W40.
| Fact | LangSmith | DeepEval / Confident AI |
|---|---|---|
| PyPI downloads, week (rank) | 18,866,051 (#2)langsmith | 686,214 (#10)deepeval |
| npm downloads, week (rank) | 8,568,634 (#1)langsmith | SDK not mapped |
| GitHub stars (rank) | 1,069 (#32)langchain-ai/langsmith-sdk | 18,682 (#7)confident-ai/deepeval |
| PyPI series (12 weeks) | ||
| npm series (12 weeks) | no npm SDK | |
| Categories | Tracing & observability, Evaluation, Prompt management | Evaluation, Tracing & observability, Prompt management |
| Status | Generally available | Generally available |
| Licence | Proprietary | Apache-2.0 (open core) |
| Self-host | Enterprise only | Enterprise only |
| OpenTelemetry | Native (OTLP) | Native (OTLP) |
| Pricing model | per-seat | flat |
| Entry price | $39/seat/mo; Plus plan per seat per month then pay-as-you-go traces (10k base traces/month included); Developer plan is $0 with 1 seat and 5k base traces | $200/mo; Confident AI platform: Free $0 forever (2 seats, 1 project, 1 GB-month traces); Starter $200/month (5 GB-months traces, then $1 per GB-month); Team $2,000/month; Enterprise custom. DeepEval library itself is free/open source. |
| Free tier | Yes | Yes |
| Measurability | Measured (mapped SDK packages) | Measured (mapped SDK packages) |
| Sources | pricing · licence · self-host · OTel (checked 2026-10-07) | pricing · licence · self-host · OTel (checked 2026-10-07) |
More downloads means more package installs (including CI and dependencies), not a better product. See the methodology.