First Impressions and Onboarding
Upon visiting the Mammouth AI site, I was struck by its bold tagline — "Don't go extinct" — and a clean, minimalist interface that immediately signals a French startup with personality. The dashboard is sparse but functional: a model selector, a chat input, and a sidebar listing available models. Signing up for the free tier (€10/month) took less than two minutes. No credit card was required for the initial trial, which lets you test the core chat experience with a limited number of messages. The app is available on Android, iPhone, and desktop, and I tested the web version on Chrome. The response speed from GPT and Claude was comparable to their native interfaces, with no noticeable lag. I did notice that the default model selection button is a bit small on mobile, but the overall feel is responsive.
Pricing and Value Proposition
Mammouth’s pricing is refreshingly straightforward. The headline plan starts at €10/month, granting access to most text and image models with a monthly message cap. For heavier users, individual model subscriptions are listed: GPT and Claude at €20/month, Mistral at €15/month, Grok at €30/month, and image generators like FLUX at €10/month. The open-source models (DeepSeek, Kimi K2.6, Qwen 3.6 Plus) have variable pricing — a potential drawback since you don't know exact costs upfront. Compared to competitors like POE or ChatHub, Mammouth offers a more curated selection and a stronger emphasis on privacy. However, the listed model versions — GPT-5.3, Claude Sonnet 4.6 — are suspiciously ahead of any official releases as of early 2025. While the site claims these versions, I could not verify them independently, which undermines trust. The €10 bundled tier is excellent for casual users who want to sample multiple AIs without paying for separate subscriptions.
Key Features and Workflow
Mammouth shines with its "One-Click Reprompting" feature. While testing, I sent a prompt to GPT-4, then with a single click sent the same prompt to Claude and Gemini. The responses were side-by-side, making comparison effortless. This multi-model workflow is the tool’s core value. Other notable features include Mammouth Projects (custom context folders), voice chat, file uploads for image and document analysis, and AI web search. The projects feature worked well for organizing a series of prompts around a client briefing. I also appreciated that all history is deletable from settings, and the GDPR compliance page details that no prompt data is retained by AI providers. One limitation: the open-source models section is sparse, with vague pricing and no clear description of how they are routed. For power users who need latest versions or fine-grained control, Mammouth may feel restrictive. The image generation quality from FLUX and Recraft was solid, though not as fast as dedicated tools.
Privacy, Limitations, and Final Verdict
Mammouth treats privacy as a first-class feature. The site clearly states that user content is never used for model training, that AI providers retain zero data, and that you can delete history at any time. This sets it apart from many free chatbot aggregators. However, the exaggerated model version numbers (e.g., GPT-5.3) are a significant red flag. Either the company is using bleeding-edge builds not publicly acknowledged, or they are misrepresenting model versions to appear cutting-edge. A senior journalist should note this caution. Also, the macOS and Windows desktop apps are not yet available, only mobile and web. Who should use Mammouth? Casual AI users who want a single, affordable hub to compare outputs from major models — and who value privacy. Power users needing the absolute latest releases or extensive API control should look elsewhere. Visit Mammouth AI at https://mammouth.ai/ to explore it yourself.
Comments