Better Stack vs Grafana Cloud vs Datadog Logs: Real Log Management Cost per 1 TB Ingested in 2026

The short answer: Datadog's architecture creates the single largest cost swing in this category, because it separates ingestion from indexing (search) into two entirely different charges, and the second one is where a 1 TB/month bill explodes. Ingesting 1 TB and simply archiving it, unindexed and unsearchable, costs $100/month on Datadog — but making that same 1 TB actually searchable for 30 days adds an indexing charge on top that can push the real bill past $12,800/month, roughly 128 times the archive-only figure. Grafana Cloud, built on Loki's label-only indexing architecture, avoids this cliff entirely: 1 TB with a month of retention costs a predictable $569/month (a $19 platform fee plus combined write/process/retain charges), because Loki never indexes full log content the way Datadog does. Better Stack's own pay-as-you-go rate card, confirmed directly, is $0.10 per GB ingested plus $0.05 per GB per month retained at its US baseline — about $150/month for the same 1 TB volume — rising with region (an EU surcharge and a Singapore rate reported as high as $0.35/GB ingested), which is likely why third-party aggregators describe a wider $0.10–$0.35 ingest range: it reflects regional pricing tiers, not a genuine rate conflict.

The billing-unit mismatch, up front

"GB ingested" does not mean the same thing across these three vendors. Datadog bills ingestion and indexing as two fully separate meters — ingestion is charged per GB regardless of whether a log is ever searched, and indexing is charged per million events, not per GB, at a rate that depends on how long that data stays searchable. Grafana Cloud (Loki) bills three separate GB-based meters — Write, Process, and Retain — that together approximate "ingest once, pay once, keep for a period," without Datadog's events-based indexing layer at all, because Loki indexes only labels/metadata rather than full log content. Better Stack bills ingestion and retention as two GB-based meters, closer in spirit to Grafana's model than to Datadog's. Comparing "price per GB" across these three without separating these components will understate Datadog's real cost dramatically and overstate how different Grafana's and Better Stack's models actually are from each other.

What each vendor bills

Datadog Log Management (OFFICIAL, Datadog's own pricing documentation, verified against the live pricing page). Ingestion: $0.10 per GB per month (annual commitment rate; $0.10 on-demand as well), charged on uncompressed data regardless of whether it is ever indexed. Logs that are ingested and archived but never indexed incur only this ingestion charge. Indexing (making logs searchable): billed per million events, by retention tier — $1.70 per million events for 15-day retention, rising to a reported $2.55 per million events at 30-day retention. Log Rehydration (pulling archived, unindexed logs back into a searchable index later) costs an additional $0.10 per compressed GB scanned. A separate Sensitive Data Scanner add-on runs $0.03–$0.04 per ingested GB, and Observability Pipelines (routing logs before they hit Datadog) adds roughly $0.095–$0.10 per ingested GB. Datadog's full pricing model spans 20-plus separately billed services beyond logs alone, widely reported as a source of unpredictable total bills.

Grafana Cloud, Logs/Loki (OFFICIAL, Grafana Labs pricing, consistent across multiple independently verified sources). Free: 50 GB/month ingested, 14-day retention. Pro: $19/month platform fee, then $0.40/GB to write, $0.05/GB to process, and $0.10/GB per month to retain — three separate per-GB meters that together cover ingest-through-retention for a given month, without a separate indexing-by-event-count charge. Loki's architectural choice to index only labels and metadata, not full log content, is explicitly what makes this flatter, more predictable cost structure possible — the tradeoff is less granular full-text search than a platform that indexes every field. Adaptive Logs, a Grafana feature that automatically drops low-value logs based on actual query patterns, was reported to cut one customer's total observability spend by 40%.

Better Stack (OFFICIAL, confirmed directly against betterstack.com's own published rate). Built on ClickHouse, bundling log management with uptime monitoring and incident management. Free: 3 GB/month, 3-day retention. Pay-as-you-go, US baseline: $0.10 per GB ingested plus $0.05 per GB per month retained — all ingested logs are immediately searchable at this flat rate, with no separate indexing tier the way Datadog has. Rates rise by region: an EU data-center surcharge adds roughly $0.05/GB to ingestion, and Singapore ingestion was reported as high as $0.35/GB — a more than 3x regional spread that any multi-region deployment should model explicitly rather than assuming the US baseline applies everywhere. Fixed bundle tiers are also available, starting around $25/month (annual) for 40 GB with 30-day retention, scaling through named tiers up to 700 GB before requiring a custom quote.

Cost per 1 TB ingested/month, by retention shape

All ILLUSTRATIVE for the Datadog event-count conversion: this article assumes 200 bytes per average log event, giving roughly 5 billion events (5,000 million) per terabyte — a reasonable midpoint for structured JSON application logs, though actual event size varies significantly by log format and verbosity.

The 128x gap between Datadog's archive-only and fully-indexed figures is the single most important structural fact in this comparison: a team that ingests logs for compliance or occasional investigation but rarely searches them pays a small fraction of what a team that needs full-text search across the same volume pays, on Datadog specifically — a distinction that doesn't exist in the same form on Grafana or Better Stack, since neither charges a separate event-based indexing fee on top of its GB-based rate.

Why Datadog's indexing charge is the real cost driver at volume

Formula: indexing cost = (bytes ÷ average bytes per event ÷ 1,000,000) × rate per million events. Because this scales with event count, not raw GB, a workload with many small, frequent log lines (a high-request-volume AI API logging a line per request) generates far more billable events per GB than one with fewer, larger log lines (a batch job logging large payloads) — meaning two teams ingesting an identical 1 TB can see dramatically different Datadog indexing bills purely based on their average log line size, a variable this article's 200-byte assumption only approximates.

Sensitivity

1. Average log event size, for Datadog specifically. Directly and inversely scales the indexing bill; smaller, more frequent log lines inflate the billable event count for the same total GB.
2. Region, on Better Stack specifically. A more than 3x swing between the US baseline ingestion rate and the reported Singapore rate; any multi-region deployment needs per-region cost modeling rather than a single flat assumption.
3. Retention length on Datadog. Moving from 15-day to 30-day indexed retention raises the per-million-event rate by roughly 50%, directly on top of the event-count multiplier.
4. Whether logs need to be indexed at all. The single largest lever on Datadog; archiving without indexing avoids the event-based charge entirely, at the cost of needing Log Rehydration (an additional per-GB charge) to search that data later if the need arises.

Budgeting traps

• Pricing Datadog off the $0.10/GB ingestion rate alone. This covers storage only; making the same data searchable is a separate, usually much larger charge billed per event, not per GB.
• Assuming "per GB ingested" means the same cost structure across all three vendors. Grafana's and Better Stack's GB-based rates already include what amounts to search capability (via label indexing or ClickHouse, respectively); Datadog's does not.
• Assuming Better Stack's US baseline rate applies in every region. EU and Asia-Pacific ingestion can run meaningfully higher, up to roughly 3.5x the US rate in the most expensive reported region.
• Ignoring average log event size when estimating Datadog's indexing cost. The same GB volume can produce wildly different indexing bills depending on how verbose or terse individual log lines are.

What to ask before you buy

Ask Datadog directly for an indexing-cost estimate based on your actual average log event size and desired retention period, not just your raw GB/month ingestion volume, since the indexing charge — not ingestion — is what actually drives cost at scale. If deploying Better Stack outside the US, confirm the exact regional surcharge directly against its pricing page before budgeting, since the gap between the cheapest and most expensive reported region is substantial.