Originally published at devtoolpicks.com
Grep and journalctl work fine right up until they do not. The moment you have more than one server, or a production incident takes an hour to debug because the relevant log line is buried somewhere across three services, you have outgrown SSHing in and searching manually. That is the actual trigger for reaching into this category, not company size or funding stage. Waiting until that point also means you skip the trap of paying for observability tooling you don't need yet, which is easy to do once vendor marketing gets involved.
Worth being clear upfront: this is not the same category as our Metabase vs Grafana vs Superset comparison. That post covers BI dashboards, turning structured data you already have into charts for business questions. This one covers raw log and trace aggregation, searching through application output to figure out what actually broke. Grafana the visualization layer can sit on top of either job, which is exactly where people mix the two up.
Quick Verdict
| Tool | Best For | Price | Rating |
|---|---|---|---|
| Axiom | Most generous free tier | Free up to 500GB/mo, Cloud from $25/mo + usage | 4.5/5 |
| Better Stack | Pure pay-as-you-go, no base fee | Usage-based from ~$0.10/GB, no subscription required | 4/5 |
| Grafana Loki | Free self-hosted, huge ecosystem | Free self-hosted, Cloud from $19/mo + usage | 4/5 |
| SigNoz | All-in-one self-hosted observability | Free self-hosted, Cloud from $49/mo | 3.5/5 |
Axiom
Axiom takes a pure usage-based approach: ingest everything, store it cost-effectively, query it with APL, its Kusto-based query language. There's no sampling and no pre-aggregation, which matters if you want to be able to search logs from six months ago without having decided in advance that they'd be worth keeping.
Confirmed from Axiom's own pricing page: the Personal plan is free permanently, no credit card required, with 500GB of data loading a month, 10 GB-hours of query compute, 25GB of storage, and 30-day retention. Axiom Cloud starts at $25 a month as a platform fee, which includes a bumped-up Always Free allowance of 1TB of loading, 100 GB-hours of query compute, and 100GB of storage, with automatic volume discounts beyond that. Enterprise add-ons like SSO, RBAC, and audit logs are self-serve at $50 to $100 a month each rather than gated behind a sales call.
For most solo projects, 500GB of free monthly ingest is more headroom than you'll actually use. The tradeoff is that Axiom is logs-and-traces focused rather than an all-in-one observability platform, so if you also want uptime monitoring or incident management, you're pairing it with something else.
Who should skip it: if you want one tool that also handles uptime checks and on-call, Axiom doesn't do that, you'll need to add a separate product. Skip it too if you specifically want a fully self-hosted option, since Axiom doesn't offer one, Personal and Cloud are both hosted by Axiom.
Better Stack
Better Stack (built from the earlier Better Uptime and Logtail products) is genuinely modular: uptime monitoring, incident management, and Telemetry (which covers logs, traces, and metrics) are priced as separate products, so you only pay for what you actually turn on.
Confirmed from Better Stack's own pricing page: the free plan includes 3GB of logs retained for 3 days. Beyond that, Telemetry access itself is free, and you pay purely for usage: ingestion starts at $0.10 per GB (annual) or $0.15 per GB (monthly), retention at $0.05 per GB per month (annual) or $0.08 (monthly). If you'd rather have a predictable number, bundles start at the Nano tier, $25 a month annual or $30 monthly, for 40GB each of logs, traces, and metrics with 30-day retention included.
The detail worth knowing: the $29-per-responder-license fee that shows up in most reviews of Better Stack is for the separate Uptime and incident management product. If you only want log search, you never touch that fee.
Who should skip it: if you want the absolute cheapest option at real scale and don't mind the operational overhead of self-hosting, Better Stack's usage-based pricing is fair but not free the way Grafana Loki's self-hosted tier is. Skip the bundles specifically if your volume is modest, pure pay-as-you-go usually works out cheaper below 40GB a month.
Grafana Loki
Loki is the log aggregation piece of Grafana's LGTM stack (Loki, Grafana, Tempo, Mimir), and it's free and open source under AGPLv3 with no license cost for self-hosting, ever. It indexes labels rather than full log content, which keeps storage cheap but means your query patterns need to lean on labels to stay fast.
Confirmed from Grafana's own pricing page: Grafana Cloud's free tier includes 50GB of logs a month with 14-day retention, community support, and access to the full Grafana Cloud stack at limited usage. Pro starts at a $19-a-month platform fee, which includes that same 50GB before pay-as-you-go kicks in, at $0.05 per GB to process, $0.40 per GB to write, and $0.10 per GB per month to retain, with retention extended to 30 days.
Self-hosting is where Loki earns its reputation as the cheapest competent option in this category, assuming you have the Kubernetes or Docker Compose experience to run the distributor, ingester, querier, and compactor components plus object storage. If you're already running other parts of the Grafana stack, or you went through the tradeoffs in our uptime monitoring roundup, the operational pattern here will feel familiar.
Who should skip it: if you want a single binary to self-host rather than several coordinated services, Loki's modular architecture is more moving parts than SigNoz's all-in-one approach. Skip Grafana Cloud specifically if 14-day retention on the free tier doesn't cover your compliance or debugging window, since getting to 30 days requires the $19 platform fee even at low volume.
SigNoz
SigNoz takes the opposite architectural bet from Loki: logs, metrics, traces, and dashboards in a single self-hosted deployment built on ClickHouse, rather than several coordinated components. It's OpenTelemetry-native, so anything already instrumented with OTel works without custom configuration.
Confirmed from SigNoz's own pricing page: the Community edition is free and open source, self-hosted, with no host or seat limits and community Slack support. Teams Cloud starts at $49 a month, which includes about $49 worth of usage (roughly 163GB of logs or traces), with additional usage billed at $0.30 per GB for logs and traces and $0.10 per million samples for metrics. Notably, Teams Cloud doesn't charge per seat or per host, unlimited team members and hosts are included in the base plan. A startup program offers the same plan at $19 a month for eligible early-stage companies.
The single-binary self-hosted story is genuinely appealing if you don't want to run four separate Grafana stack components. The tradeoff is that SigNoz Teams Cloud's $49 base fee is a bigger jump from free than Better Stack's zero-base-fee model or Axiom's generous free tier, so it only makes sense once you're actually using a meaningful chunk of that included usage.
Who should skip it: if your usage is modest and you're on the cloud plan rather than self-hosting, the $49 base fee is harder to justify than Axiom's free 500GB or Better Stack's pure pay-as-you-go pricing. Skip self-hosting Community edition too if you don't want to operate ClickHouse yourself, it's the same operational tradeoff as self-hosting Loki, just with one service instead of four.
How Do You Actually Choose Between These?
Start by asking whether you need this at all. If your app runs on one server and you're still comfortable running journalctl -u your-app over SSH when something breaks, you don't need any of these yet, and adding one now is solving a problem you don't have. Once you cross into multiple services or you're regularly hunting for a log line from more than a day ago, pick based on your actual monthly volume. In a worked example of 15GB of logs a month with 30-day retention, Axiom's free tier covers it entirely at $0, Better Stack's pure pay-as-you-go pricing lands around $3.45, self-hosting Loki or SigNoz costs roughly the price of a small VPS, Grafana Cloud Pro runs $19 for the platform fee alone, and SigNoz Teams Cloud comes in at $49. Under a few hundred GB, the free and pay-as-you-go options are the ones worth starting with. If you'd rather have logs, traces, and error context in one place, our Sentry alternatives roundup covers the error-tracking side of this same decision.
Which Log Management Tool Should You Actually Use?
For most solo developers just crossing the "I need real log search" threshold, start with Axiom's free tier, 500GB a month covers nearly everyone at this stage and there's no reason to pay before you have to. If you want the most predictable, transparent pay-as-you-go pricing with zero base fee, Better Stack is the pick, especially if you'll eventually want uptime monitoring from the same vendor. Choose self-hosted Grafana Loki if you're already in the Grafana ecosystem, or SigNoz Community if you'd rather run one binary instead of four. Skip SigNoz's and Grafana Cloud's paid tiers specifically until your usage actually approaches their included allowances, paying a base fee for headroom you're not using yet is the one mistake to avoid here.
Top comments (0)