Video: "Minimax 2.1 + Claude 4.5 + Gemini 3.0 SEO is INSANE!" by Julian Goldie on YouTube.
What Minimax 2.1 actually is
Minimax is a Chinese AI company, and their M2.1 model is currently available free on several platforms including OpenRouter and their own interface. The headline number that's been turning heads is context length — Minimax 2.1 can hold a substantial body of content in a single session, which matters for SEO work where you're often feeding in site data, competitor pages, and keyword lists simultaneously.
The model has scored well on structured reasoning benchmarks, which is exactly the kind of task that keyword clustering and content outline generation requires. That said, benchmarks measure performance on specific test sets — they don't always translate into consistent real-world output. Worth knowing: as a free, non-European model, Minimax's data handling terms deserve a look before you feed in anything client-sensitive or commercially confidential.
How the three compared on SEO tasks
Julian ran all three on the same brief: take a seed keyword in a specific niche, produce a full keyword cluster with search intent labels, draft a content outline for the primary term, and write three meta descriptions at different lengths. The outputs were compared side by side.
On keyword clustering, Minimax was fast and grouped logically — the intent labelling was consistent and the structure was easy to work with. Claude produced fewer clusters but added a layer of reasoning about which terms were genuinely commercial versus informational, which is the kind of nuance that changes where you spend your content budget. Gemini's clusters were structured similarly to Minimax but with slightly more padding — useful as a starting point, needing more editing before it's actionable.
On content outlines, Claude pulled ahead. The section structure was better thought through, the H2 suggestions matched real user questions rather than keyword-stuffed headings, and the flow read like a human had designed it. Minimax produced a competent outline; Gemini's was solid but generic.
Meta descriptions: Claude again. Minimax hit the length requirements but the copy was functional rather than compelling. Gemini tended to write slightly long and needed trimming. For volume production of metas where quality is "good enough", Minimax is a viable free option.
The multi-model approach — using each where it's strongest
Julian's takeaway wasn't "use this one model for everything" — it was that different models suit different parts of an SEO workflow. Minimax for initial clustering and bulk intent labelling. Claude for content strategy, outline architecture, and anything that needs genuine editorial judgement. Gemini for cross-referencing Google's own search patterns and producing content that mirrors how Google structures answers.
This is how experienced AI SEO practitioners are approaching the stack now. You don't pick a model; you pick a task and pick the right tool for it. In practice, that means having API access to a couple of models rather than relying on a single provider — which does add cost and setup complexity.
What this means for a UK business doing SEO
For most businesses, running a multi-model stack is overkill. If you're producing a handful of pages per month, Claude 4.5 or Gemini 3 via a single subscription covers the use case well enough. The multi-model approach starts paying off when you're producing content at scale — dozens of pages per month, across multiple topic clusters — and the difference between "competent but generic" and "editorially strong" affects rankings.
That said, the Minimax finding is worth noting: a free model that performs competently on SEO clustering is a genuine option for initial keyword work, even if you then pass the output to Claude for refinement. It won't replace a proper SEO process, but it compresses the time spent on the less creative parts of it.
Where this connects to NordSys
We run SEO and Generative Engine Optimisation for UK businesses, building content strategies that work both in traditional search and in AI answers from Google, ChatGPT and Perplexity. If you want to know which AI model is right for your SEO workflow — or whether AI belongs in your content process at all — that's a conversation worth having before you commit to a stack.
See our SEO & AI Ranking service →