OBSERVE

OBSERVE

How Observe pricing works

Everything that shows up on your bill, what drives cost up or down, and example scenarios so you can estimate before you commit.

Primary unit: agent executions / month

All plans: unlimited users

Cost components

Cost components

What's on the bill

Every possible line item for Observe, explained. No hidden fees.

Agent executions

PRIMARY METRIC

The core unit. One execution = one top-level agent run, regardless of how many LLM calls, tool invocations, or spans happen inside it. Sub-agents spawned by a parent each count as their own execution.

EXPLORER (FREE)

[X]K / month

Pauses at limit — no bill

pro (included)

[X]K / month

Overage: $[X] per 1K

ENTERPRISE

Custom

Negotiated volume + rate

Platform Fee

PRO + ENTERPRISE

The base monthly fee for Pro covers access to the full platform — SSO, RBAC, audit logging, and the dev loop (prompts, datasets, evaluators, experiments, provider routing). Explorer has no platform fee.

EXPLORER (FREE)

$0

pro

$[X]K / month

Monthly or annual billing

ENTERPRISE

Custom

Annual contract

Execution packs

optional

Pre-purchase execution volume at a discount instead of paying overage rates. Available on Pro. Unused executions don't roll over. Contact sales for pack sizes and pricing.

Data residency

optional - Enterprise

Choose where your trace data is stored. Available as a paid add-on for Pro. Included at no extra cost on Enterprise (US East or EU West).

What affects your bill

Cost drivers

What makes your Observe bill go up or down so you can optimize before you scale.

increases cost

Sub-agent spawning

If your coordinator spawns 5 sub-agents per run, that's 6 executions (1 parent + 5 children). Consolidating logic into fewer agents reduces execution count.


increases cost

High-frequency agents

Agents that run on a cron (every minute, every 5 minutes) accumulate executions fast. A per-minute agent = ~43K executions/month. Use scheduling policies to throttle.

increases cost

Extended retention

Enterprise retention (up to 3 years) is more expensive than standard. If you don't need long retention for compliance, shorter windows keep costs lower.


decreases cost

Batching within a run

Multiple LLM calls, tool invocations, and retrieval steps within one agent run all count as a single execution. More work per run = better unit economics.

decreases cost

Execution packs

Pre-purchasing volume is cheaper than overage rates. If you know your monthly volume, packs lock in a lower per-execution price.


decreases cost

Annual billing

Annual plans reduce the platform fee. The execution rate stays the same, but the base cost drops.



Your Observe bill

How the receipt adds up

Every line item maps back to a cost component above. No hidden charges this is all of it.

Cost calculator

Traced executions / mo10Kexecutions
030K
Team seats2seats
12
Platform fee$0
10K executions← included$0
OveragePaused at cap
Data retention14 days$0
Monthly total$0
Pauses at limit — no surprise bills. Upgrade to Team or Business anytime.

FAQ

What counts as an "agent execution"?

One agent execution = one top-level agent run. All nested spans, model calls, and tool uses within that run are included at no additional cost. Think of it as one "conversation" between your agent and the world — however many LLM calls it takes to complete. If your agent spawns sub-agents, each sub-agent run counts as its own execution.

Why per-tenant instead of per-seat?

Observability and governance tools need wide adoption to be useful. Per-seat pricing discourages teams from giving everyone access, which defeats the purpose. We charge per tenant so your entire team — developers, ops, compliance, leadership — can see what your agents are doing without anyone doing headcount math.

What's the difference between Observe and Runtime?

Observe instruments your agents wherever they run — your AWS, your Kubernetes, your laptop. It gives you visibility and policy enforcement as a library. Runtime is the execution environment itself: it runs your agents on Waxell's infrastructure with durable workflows, pre-execution policy gates, isolated compute, and kill switches. Most teams start with Observe and add Runtime when the stakes of their workflows demand it — financial automation, healthcare, infrastructure changes.

Can I use just one product, or do I need the whole platform?

Use what you need. Many teams start with just Observe or just the MCP Gateway. Each product works independently with its own pricing. The platform fee on Pro covers access to everything — you only pay usage on the products you actually use. There's no penalty for starting small.

How does the Startup Program work?

Apply with your company details. If you're Series A or earlier, under [X] employees, under $[X]M in funding, and building with AI agents, you qualify for deeply discounted Pro-tier access for [X] months. You get the full platform — not a watered-down version. When you graduate, you move to standard Pro pricing. The idea is simple: governance should be a Day 1 decision, not a Series B retrofit.

Is there a commitment? Can I cancel anytime?

Monthly plans have no commitment — cancel anytime. Annual plans are billed upfront at a 20% discount. Enterprise plans have custom terms. Explorer is free forever with no strings attached.

What happens if I go over my included executions?

On Explorer, execution beyond your limit pauses until next month (no surprise bills — ever). On Pro, overage is billed at a transparent per-1K rate — you'll see it in real time on your dashboard. We'll alert you at 80% and 100% of your included volume. You can also pre-purchase execution packs at a volume discount.

Waxell

Waxell provides observability and governance for AI agents in production. Bring your own framework.

© 2026 Waxell. All rights reserved.

Patent Pending.

Waxell

Waxell provides observability and governance for AI agents in production. Bring your own framework.

© 2026 Waxell. All rights reserved.

Patent Pending.

Waxell

Waxell provides observability and governance for AI agents in production. Bring your own framework.

© 2026 Waxell. All rights reserved.

Patent Pending.

Waxell

Waxell provides observability and governance for AI agents in production. Bring your own framework.

© 2026 Waxell. All rights reserved.

Patent Pending.

1 part time agent