<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
 <title>TestedActually newsletter</title>
 <link>https://testedactually.com</link>
 <description>New comparisons, head-to-heads, and original benchmark data from TestedActually, with the date every claim was last checked.</description>
 <language>en</language>
 <lastBuildDate>Tue, 06 Oct 2026 00:00:00 GMT</lastBuildDate>
 <atom:link href="https://testedactually.com/newsletter/rss.xml" rel="self" type="application/rss+xml" />
 <item>
  <title>Issue 003: Twelve GTM tools benchmarked, and Mixmax drops its cheapest tier</title>
  <link>https://testedactually.com/newsletter/2026-10-06-gtm-expansion-and-market-events/</link>
  <guid isPermaLink="true">https://testedactually.com/newsletter/2026-10-06-gtm-expansion-and-market-events/</guid>
  <pubDate>Tue, 06 Oct 2026 00:00:00 GMT</pubDate>
  <description>Our biggest ship yet: the GTM expansion, two new clusters covering the CRM and marketing-automation stack with crawler benchmarks and head-to-heads for the highest-demand pairs. Alongside it, pricing guides for opencode and Fathom and for the local-LLM desktop apps, a Bland voice-AI ladder we verified to the enterprise tier, and six dated market events from the news scan, including Mixmax's entry price moving from $29 to $49 a user.

* Ollama pricing, free local models plus a $20 cloud tier (Comparison): The local-LLM staple keeps local use free forever; the cloud plans run Pro at $20 a month ($16.67 billed annually, $60 of monthly usage credits), Max at $100 ($300 of credits), and Team at $500 ($1,000 of shared credits) [1]. Off-peak rates and per-credit details are in the guide.
* Bland — pricing, features, and fit (Comparison): Bland repositioned upmarket this year, so every claim is re-sourced to its live site and the full ladder is public: Start is free with call time at $0.14 a minute, Build drops that to $0.12 behind a $299 monthly platform fee, Scale runs $0.11 at $499 a month, and Enterprise is contracted to your volume [2].
* Best CRM platforms for startups (HubSpot, Pipedrive, Attio, Close, folk, Zoho) (Comparison): Six CRM platforms compared on visual pipelines, native email sync, bundled marketing, and free tiers. HubSpot is the platform default with a real free tier (2 users, 1,000 contacts); Pipedrive owns the single-pipeline job from US$14 per seat; Attio bends its data model to your business. HubSpot Professional's $3,000 mandatory onboarding fee is on its own pricing page.
* Best marketing automation platforms (Klaviyo, Omnisend, Customer.io, ActiveCampaign, Mailchimp, Braze) (Comparison): Six marketing automation and lifecycle platforms split by where their triggers live: storefront (Klaviyo, Omnisend, Mailchimp), product events (Customer.io), builder-first (ActiveCampaign), or enterprise cross-channel (Braze). Free tiers: three are real, one of those has no automation flows, and Braze publishes no prices at all.
* AI crawler access test: can answer engines read these vendors? (Original data): Six CRM vendors probed three times each on 2026-09-30. Attio and folk block all seven AI crawlers while selling themselves as the CRM for agentic revenue; HubSpot, Pipedrive, Close, and Zoho are fully open. Five of six ship an llms.txt.
* AI crawler access test: can answer engines read these vendors? (Original data): The whole marketing-automation category is open to AI crawlers: zero robots blocks across all six vendors, llms.txt on five of six. If LLMs under-recommend a tool in this category, it is not a robots problem.
* Pipedrive vs HubSpot — pipeline or platform (Head to head): Pipeline or platform. Pipedrive's bill grows with seats (US$14 to start); HubSpot's grows with seats, contacts, hubs, and onboarding fees, and its free email sync is the zero-cost logging layer Pipedrive only matches at US$39 per seat.
* Klaviyo vs Mailchimp — ecommerce engine or email app (Head to head): The entry bands tie at $20/mo, then diverge on what is inside: Klaviyo's free plan runs flows and includes $5 of mobile messages, Mailchimp's free plan includes no automation flows at all. Mailchimp's own homepage advertises switching savings of up to 42 percent, which tells you who it is fighting.
* Jan vs LM Studio — the open-source app or the model workbench (Head to head): The two open-source desktop apps for running local models, compared on licence, interface, model management, and API serving. Jan if you want a free MIT-licensed app that stays out of the way; LM Studio if you want the deeper workbench and its bundled Bionic agent, and will confirm the workplace licence term first.
* opencode pricing, free software with a pay-per-request model gateway (Comparison): opencode costs nothing to install or run: the agent is free and open source and connects to 75+ model providers you bring yourself. The only money path is Zen, an optional model gateway billing per request at provider rates with the markup capped at processing fees.
* Fathom pricing, a real free plan and paid tiers from $15 a user (Comparison): The meeting notetaker with a free plan that works rather than teases: unlimited recordings, transcriptions, and AI summaries for individuals. Paid tiers add depth from $15 a user (Team, annual) to $20 a month Premium, and the guide marks where the free plan genuinely ends.
* Mixmax retires its SMB tier, leaving Copilot and Suite only (News): Mixmax's pricing page now lists Copilot at $49 and Suite at $89 per user per month, and no longer offers the $29 SMB tier that appeared in our 2026-09-24 check. The cheapest listed path into paid Mixmax is now $49 a user.
* Aleph Alpha ships Kolibri: Apache 2.0 open-weight MoE with 1M-token context (News): Aleph Alpha released Kolibri, an English-German Mixture-of-Experts model with 78.1B total parameters (3.46B active), full weights on Hugging Face under Apache 2.0, and 262K native context extending to 1M via a serve-time flag. A permissive, self-hostable option for EU-data-residency stacks.
* DeepKeep ships AI Lens: a security control plane for coding agents (News): DeepKeep announced AI Lens for Developers: policy enforcement, audit visibility, and runtime security over coding agents, supporting Cursor and Claude Code today. The coding-agent security lane is forming fast, two weeks after StackHawk's Wingman landed at $10 per user per month.
* Underdog launches a fully local personal AI agent, no cloud (News): Underdog announced a fully on-device personal AI agent: it reads email, takes meeting notes, handles dictation, manages the calendar, and drives the browser, asking before it pays, with nothing uploaded. Invite-only beta, no pricing disclosed.
* Deepslate raises €7.7M seed for European speech-to-speech models (News): Berlin voice-AI lab Deepslate raised a EUR 7.7 million seed led by 42CAP, claiming 440 ms speech-to-speech latency. More model choice and a potential EU-hosted vendor for voice-agent platform buyers at a time when data residency is a real procurement question.</description>
 </item>
 <item>
  <title>Issue 002: Reclaim's pricing ladder verified, and Voiceflow stops publishing numbers</title>
  <link>https://testedactually.com/newsletter/2026-09-28-reclaim-and-voiceflow-pricing/</link>
  <guid isPermaLink="true">https://testedactually.com/newsletter/2026-09-28-reclaim-and-voiceflow-pricing/</guid>
  <pubDate>Mon, 28 Sep 2026 00:00:00 GMT</pubDate>
  <description>Two new pricing guides: Reclaim, from a free Lite plan to $22 a seat, and Voiceflow, which no longer publishes tier prices at all. Also this week: our Netlify free-plan facts are corrected against Netlify's own billing docs, and Mailmeteor's Outlook support is now documented.

* Reclaim pricing and review, free calendar defence from $12 a seat (Comparison): The budget pick for AI calendar defence: Lite is free forever for a solo user, Starter is $12 a seat monthly or $10 annual, Business $18, and Enterprise $22 annual [1]. Pay Motion's prices only if your task list needs sequencing too. Ladder re-verified on the vendor pricing page, 28 September 2026.
* Voiceflow pricing and review, governance for customer-facing chat agents (Comparison): Still our pick for governed, customer-facing chat agents, but as of 28 September 2026 the vendor publishes no tier numbers: a free trial without a credit card, usage-based billing, and a sales conversation for business [2]. Botpress lists $150 a month for Plus if you need a published figure to budget against [3].</description>
 </item>
 <item>
  <title>Issue 001: Six free hosts tested, and the crawlers vendors actually block</title>
  <link>https://testedactually.com/newsletter/2026-09-24-free-hosts-and-crawler-access/</link>
  <guid isPermaLink="true">https://testedactually.com/newsletter/2026-09-24-free-hosts-and-crawler-access/</guid>
  <pubDate>Thu, 24 Sep 2026 00:00:00 GMT</pubDate>
  <description>The first issue: our free-hosting benchmark, the voice-agent head to head, and the crawler matrix asking which AI vendors let answer engines read their sites.

* Best free web hosting for SPAs (tested) (Comparison): Cloudflare Pages is the default free host for most SPAs: unlimited bandwidth, functions included, no trial clock. Vercel wins for non-commercial Next.js, Netlify for workflow extras.
* AI crawler access test: can answer engines read these vendors? (Original data): Six free hosts, probed on 2026-09-20: median TTFB, robots policy for AI crawlers, and who ships an llms.txt. GitHub Pages answered fastest at 21ms median.
* Vapi vs Retell — raw control or a builder that ships (Head to head): Raw API control or a builder that ships: which team shape fits Vapi, which fits Retell, judged on inbound and outbound calling, campaign dialing, and whether a no-code builder exists.
* Research: how AI companies treat each other's crawlers (Research note): Eight AI vendors against eight answer-engine crawlers, robots policy and live status per cell, from a datacenter vantage. OpenAI serves GPTBot a 200 while 403-challenging ChatGPT-User, its own fetcher.</description>
 </item>
</channel>
</rss>