Best cold email infrastructure · inbox providers

Matthew Lucero's team ran the public grid most listicles still skip. Over Q4 2025 into Q1 2026 they sent 650,075 cold emails split evenly across Instantly, PlusVibe, and Smartlead, and across Hypertide, Inboxkit, and Zapmail. Same lead lists, same copy, same warmup periods, same sending windows.

Human reply landed at Hypertide 0.24%, Inboxkit 0.22%, and Zapmail 0.22%. At 95% confidence those provider gaps were not meaningful. Sequencer human-reply gaps in the same grid were small too. List quality, offer, copy, hygiene, and follow-up speed still outrank inbox brand.

So the shortlist is a service-style picker:

• Premium Inboxes when you want done-for-you official Google or Microsoft with replacements.
• Inboxkit when API, Azure, and scripted fleets are the job.
• Zapmail when pre-warmed speed matters and you already priced year two.
• Maildoso or Mission Inbox when SMTP or dedicated IP sits under your sequencer.
• ScaledMail when one managed vendor across Google, Outlook, and SMTP is the ask.
• Hypertide when cheap Microsoft density on Microsoft 365 is the cost problem.
• Direct Google or Microsoft when control on brand-primary domains beats reseller speed.

If someone asks which inbox brand wins on reply rate among those three Lucero tested, the honest answer is none of them at that size.

Frederik Jakobsen at Danish Lead Co published a different kind of proof. Between April and July 2026 their agency sent 1,570,049 emails across 442 campaigns for 124 active clients, tracked at the individual provider level. It is observational, not a shared-list lab. The lesson still lands.

Provider-level bounce can hide inside a healthy campaign average. Google Zapmail sat near the top of their Google rows on reply in that book. Outlook Hypertide sent 453,084 messages at 0.87% reply and 5.93% bounce. That bounce is about four to five times their best Google rows. Cheap Microsoft density on Microsoft 365 is real. So is the need to chart bounce by provider, not only by campaign.

Receiving-side gateways set another ceiling. In the same Danish writeup, positive reply by prospect gateway went near zero on enterprise security stacks. Proofpoint and Mimecast showed 0.00% positive reply in that table. Google Workspace prospects sat higher. Microsoft 365 sat near the floor. If your buyers live behind those gateways, swapping Zapmail for Premium will not invent replies.

Bounce is also not a reply promise. Danish's Inboxlee row showed healthy bounce with weak reply versus their Zapmail Google lane. Low bounce means the pipe is open. Messaging and targeting still decide whether you hit the ceiling.

Lucero's human-reply tie and Danish's observational Zapmail Google lead can both be true. Different metrics, different designs. Label them. Do not mash them into one crown.

Year-one stickers lie if you stop at month twelve. Zapmail's help center publishes what happens after 12 months active on Google and Microsoft mailboxes. Starter steps to $59 plus $6 per mailbox. Growth steps to $169 plus $5.50. Pro steps to $499 plus $5.

Work a concrete Pro example from those published numbers. One hundred Pro mailboxes on yearly plan is $3,000 a year, about $250/mo list in year one. Year-two Pro list math is $499 plus 100 times $5, which is $999/mo. That is about four times the year-one annual monthly. Listicles still quote the year-one starts-at language.

Premium Inboxes and Inboxkit do not publish an equivalent auto cliff in the same help-doc shape. Still ask how they pass through native Google and Microsoft license renewals. Direct Workspace and Microsoft seats have the same class of renewal problem once you add labor.

Pick Zapmail when you need inboxes this week and you have already priced month 13. Pick Premium or Inboxkit when you are modeling a year of production, not a burst. Pre-warmed speed is a time trade, not a forever placement guarantee.

Official Google Workspace and Microsoft tenants and SMTP cubes are different jobs. Premium, Inboxkit, and Zapmail sell official-tenant paths (plus Azure or Microsoft 365 mailbox volume on Inboxkit and Hypertide). Maildoso and Mission Inbox sell cheap SMTP send cost and dedicated IP under your sequencer.

Cheap SMTP wins unit economics conversations. It loses when your team refuses anything but Admin Console logos, or when the job is production reply into Google Workspace buyers. Danish's near-zero reply SMTP example is a product-fit warning for the wrong job. A Reddit head-to-head that put Maildoso and Inboxkit near 2% reply on 20k sends after buying 50 accounts each is a different test shape. Keep those stories separate.

ScaledMail sits in the middle with published Google, Outlook, and SMTP packages on one vendor. Hypertide Microsoft 365 at $25 per domain fills Microsoft density when per-mailbox pricing hurts, with the Danish Outlook bounce caution attached.

Rule of thumb: diversify Google and Microsoft. Do not put everything on one host. Keep SMTP as a lane, not the only lane, unless that is explicitly the product you sell.

A few names show up in the same conversations and still do not earn a ranked slot here.

• Warmy is warmup and monitoring, not an inbox provider.
• Instantly done-for-you stays out as infra. Keep Instantly as the sequencer.
• Saleshandy, Lemlist, and Apollo infra blur sequencer and mailbox layers. Wrong primary shortlist for this page.
• Mailforge, Primeforge, and similar listicle fixtures are peer name-drops, not this catalog's production shortlist.
• Puzzle Inbox, Cheapinboxes, Peeker, Icemail, and Inboxlee stay long-tail or observational footnotes.

How to pick this week:

Figure out what is breaking first. If mailboxes keep dying and you need someone else to replace them, open Premium Inboxes. If you want to spin up lots of boxes yourself with an API, open Inboxkit. If you need boxes this week and you already checked next year's price, Zapmail can work. Buy 20 to 50 on the same list and the same copy before you lock a year. Watch bounce and replies by provider every week.