Dash0 Review (2026): Pricing, Features & Honest Verdict

TLDR

Dash0 is an OpenTelemetry native observability platform covering logs, metrics, traces and profiles. It bills per signal rather than per host or per seat, so the invoice finally reads like arithmetic. Best for: cloud native engineering teams already emitting OpenTelemetry. Price: usage based from $0.20 per million metric data points, no free plan. Rating: 8.1/10

What is Dash0?

Dash0 is a hosted observability backend built around OpenTelemetry instead of a proprietary agent. You point an OpenTelemetry collector at it and it ingests metrics, log records, spans, span events, real user monitoring events, synthetic check results and continuous profiles. Everything lands in one store called SignalStore. You query it with PromQL or SQL, build dashboards and alerts as code, and reach the whole thing through a CLI, an API or an MCP server if you want an agent doing the digging.

What that buys you in practice is one place to answer the question every incident starts with: what changed. Service maps show you the topology, traces show you the slow path, logs show you the error, and profiles now show you the exact function burning the CPU. We found the interface unusually fast to move around because it is keyboard driven, which sounds like a small thing until you are three minutes into an outage and clicking through nested menus.

The company behind it is Dash0 Inc., founded by Mirko Novakovic, who previously co-founded Instana before IBM bought it. That matters, because Dash0 is essentially Novakovic’s second attempt at the same problem with the vendor lock-in removed. Dash0 raised a $110 million Series B at a $1 billion valuation, and in August 2026 it acquired Polar Signals, picking up the best continuous profiler on the market plus a storage engine called Great Lakes that now replaces ClickHouse underneath SignalStore. The timing is not an accident. OpenTelemetry has become the default way to instrument software, which quietly destroyed the moat that agent based vendors spent a decade building. The fight is now about query speed and the size of the bill.

What Are Dash0’s Key Features?

OpenTelemetry Native Ingestion

There is no Dash0 agent to install. Anything that speaks OTLP works, which means your existing collector config, your existing SDK instrumentation and your existing Prometheus scrapes port over without a rewrite. We dug into the signal coverage and it is genuinely broad: infrastructure and Kubernetes monitoring, APM, distributed tracing, log management, website monitoring, synthetic checks, service maps and alerting all sit in the single plan. The catch is retention, which is fixed per signal type rather than configurable. Metrics and synthetic check runs are kept 13 months. Spans, span events, log records and web events are kept 30 days. If your compliance team wants a year of raw logs, that is a conversation you need to have before you sign anything.

Keyboard Driven UI and Config as Code

Dashboards and alert rules are defined as code and can live in your repo alongside everything else, which means review, rollback and templating across environments all work the way your team already works. Queries run in PromQL or SQL depending on what you are asking, so nobody has to learn a bespoke query language invented by a vendor. The interface is built for speed rather than screenshots, with keyboard navigation throughout and a dark mode that was clearly the default rather than an afterthought.

Per Signal Pricing With Budget Controls

Dash0 charges for what you send and nothing else. No per seat fee, no base platform fee, no feature gating between tiers, because there are no tiers. Every account gets every feature. Alongside that, Dash0 ships monthly budget limits, spam filters for runaway telemetry and cost forecasting, which are the three things every finance team asks for after their first surprise observability invoice. Being able to hard cap a month is unusual in this category and it is one of the strongest arguments for the product.

Agent0 and AI Coding Insights

Agent0 is Dash0’s AI site reliability engineer. It triages incidents against your telemetry and proposes production fixes, billed at $0.60 per credit rather than folded into a seat price. Separately, Darkplane and AI Coding Insights track what your AI coding tools are actually shipping and what that code does in production, at $10 per user per month. This is the newest and least proven part of the platform. We would treat it as a bonus rather than a reason to buy, and validate it during the trial before budgeting for it.

Continuous Profiling Including GPU

The Polar Signals acquisition brings line level CPU and memory profiling into the same platform as your traces, plus production safe GPU profiling for NVIDIA CUDA workloads across both inference and training. Almost nobody else does GPU profiling safely in production, and if you are running model serving on expensive accelerators that single capability can pay for the subscription. Profile data also feeds Agent0 over MCP, so the AI side has code level context rather than just span timings.

How Much Does Dash0 Cost?

Dash0 has exactly one plan and it starts at $0.20 per million metric data points. Spans, span events, log records and web events are $0.60 per million. Synthetic API check runs are $0.20 per thousand. Agent0 credits are $0.60 each and AI Coding Insights is $10 per user per month. There is no base fee and no seat fee. The trial is 14 days with full feature access, Agent0 credits included and no credit card required, but there is no free forever tier once the trial ends.

Against the incumbents this reads well. Datadog charges $15 per host per month for Infrastructure Pro on annual billing, another $31 per host per month for APM, and then $0.10 per ingested GB plus $1.70 per million indexed log events on top. Grafana Cloud Pro adds a $19 monthly platform fee plus $6.50 per 1,000 billable metric series and roughly $0.45 per GB to ingest logs. SigNoz Cloud carries a $49 per month minimum with logs and traces at $0.30 per GB on 15 day retention. Dash0 removes both the per host tax and the per seat tax, which is where most observability bills quietly triple.

The honest caveat: per record pricing rewards disciplined instrumentation and punishes chatty debug logging. A service emitting 500 million log records in a month costs $300 at Dash0 before you have monitored a single metric. That is fine if your volumes are known, and painful if a bad deploy starts looping an error log. The spam filters and budget caps exist precisely because that scenario is common. Model your real monthly signal volumes during the trial. Do not sign based on the headline rate.

Who is Dash0 Best For?

Use Dash0 if you are already instrumented with OpenTelemetry and tired of doing algebra to explain your monitoring bill. It is a strong fit for platform teams running Kubernetes and microservices, for engineering orgs where everyone needs access but nobody wants to pay per seat for it, and for anyone running GPU workloads who wants profiling and tracing in the same pane. The config as code approach for dashboards and alerts also suits teams that already keep infrastructure in version control.

Skip Dash0 if any of these apply:

• You need a free forever tier. Grafana Cloud and self hosted SigNoz both give you one, Dash0 gives you 14 days.
• You depend on Datadog’s hundreds of turnkey integrations, security monitoring or CI visibility. Dash0’s surface area is narrower by design.
• You need log or trace retention beyond 30 days without exporting elsewhere.
• Your stack is not instrumented with OpenTelemetry and you have no appetite for that migration. The whole value proposition assumes you have done that work.

One more consideration: Dash0 is a young company with a lot of fresh capital and a founder who has sold in this category before. That cuts both ways. You get rapid feature shipping and a team that clearly understands the problem, but you are also betting on a platform that has not yet been through a decade of enterprise procurement, compliance audits and edge case bug reports the way Datadog has. Run a real workload through the trial rather than a toy service.

Best Dash0 Alternatives

Grafana Cloud

The obvious comparison if you already live in the Prometheus, Loki and Tempo world. Grafana Cloud has a genuinely free forever tier, and Pro costs a $19 monthly platform fee plus $6.50 per 1,000 billable active series and roughly $0.45 per GB for log ingest. It is cheaper at small scale and it wins on ecosystem breadth. It loses on coherence: you are assembling four products with four billing models, and active series pricing gets ugly the moment Kubernetes labels blow up your cardinality. Pick Grafana if you want flexibility, pick Dash0 if you want one bill and one query surface.

SigNoz

The open source option, and the closest philosophical match since SigNoz is also OpenTelemetry native. You can self host the whole thing for the cost of running it, or use SigNoz Cloud which has a $49 per month minimum with logs and traces at $0.30 per GB on 15 day retention, $0.40 per GB at 30 days and $0.10 per million metric samples. SigNoz is less polished than Dash0 and has nothing comparable to Agent0 or GPU profiling, but if you want the data on your own infrastructure with no vendor at all, it is the answer.

Datadog

The category leader and the tool most teams are trying to escape. Infrastructure Pro is $15 per host per month on annual billing, APM adds $31 per host, and logs run $0.10 per ingested GB plus $1.70 per million indexed events. Nothing matches its integration catalogue, its security products or its maturity at enterprise scale. Nothing matches its capacity to produce a bill nobody in the room can explain either. If you need the breadth, pay for Datadog. If you mostly need logs, metrics, traces and profiles, Dash0 does that job for a fraction of the cost.

Final Verdict: Is Dash0 Worth It?

Dash0 is the most credible OpenTelemetry native observability platform we have looked at this year. The pricing model is the headline feature: charging per signal, with no seats and no hosts and a hard monthly cap, removes the single biggest complaint engineering leaders have about this category. The Polar Signals acquisition turned a good tracing and logging product into one with genuine depth, and production safe GPU profiling is a real differentiator that nobody else is offering at this price.

It earns 8.1 rather than higher for two reasons. Fixed 30 day retention on logs and traces is inflexible, and the absence of any free tier makes it a harder sell to small teams who can start on Grafana Cloud for nothing. Agent0 is also too new to bank on. But if you are running production services on OpenTelemetry and your Datadog renewal is coming up, put Dash0 on the shortlist and run your real telemetry volumes through the 14 day trial. The arithmetic will probably surprise you.