Skip to content
PricingSaaS
Sign in
Back
Pricing briefing
PPerplexity
Perplexity
PACKAGING CHANGE

Perplexity swapped Gemini Flash for Gemini Hydrogen

By , Co-founder, PricingSaaS

Published

Perplexity swapped Gemini 3.8 Flash for Google's new Gemini 4 Hydrogen model.

See before and after

Perplexity’s pricing page · Change detected

What changed

Perplexity swapped the Gemini entry in its multi-model lineup: Gemini 3.8 Flash — described on the page as Google's fast model — was removed and replaced with Gemini 4 Hydrogen, described as Google's latest model and marked with a "New" badge. The swap touches only that one row in the model-selection card; the rest of the lineup is untouched and no plan price changed.

Why it matters

This is a packaging move, not a pricing one: Perplexity didn't touch what any plan costs, it changed which model answers a query under its Pro and Computer plans. For a product leader watching the AI-bundling arc, that's the real signal — model curation, not model pricing, is becoming a lever companies pull to keep a bundled-AI product feeling current. Pulse has tracked 11 model-lineup changes from Perplexity in the last 365 days — about one every 12 days.

Swapping in a model tagged "New" costs Perplexity nothing in margin and nothing in price friction, but it resets how fresh the product looks to anyone comparing AI tools side by side. If you bundle third-party models or capabilities into your own packaging, Perplexity's refresh cadence is worth benchmarking: watch whether your own lineup updates are keeping pace with what users can see a competitor ship for free.

In context

Perplexity has made nine changes to this same model-lineup card since mid-July, per Pulse's tracking — including a same-week reversal, swapping in Grok 4.7 for Claude Haiku 5.5 on Oct 9 then flipping back the next day to Claude Haiku 5.5 over Grok 4.7, and upgrading Claude Sonnet 5 to 5.5 on Sep 29. That pattern reads less like a settled roster and more like a dial Perplexity turns often. The 36 days since its last change sits well above Perplexity's own 12-day average cadence, per Pulse's tracking since Jul 2026 — a longer pause than usual for this company. Across the market, Pulse tracked 329 feature-type changes among 269 companies, including HubSpot, Slack and Cognition, in the last 180 days, so lineup churn like this is common; the sharper read is that showing off the newest foundation model, not its price, is becoming the feature SaaS-wrapped AI products compete on.

Common questions

Is this a one-off or part of a pattern?

Per Pulse, Perplexity touches this card roughly every 12 days on average, and is tied with HubSpot for the most feature-type changes among tracked companies — six each — over the last 180 days. What the record doesn't show is whether usage data or each model vendor's own release calendar is driving the swaps.

Do competitors change their AI model lineups this quietly, without a price change?

Not always — on Nov 11, 2026, Slack shifted Slackbot to a 15-message weekly usage limit and Cognition halved its Team plan's user cap, both pairing a product change with a pricing or limits move, per Pulse. Perplexity's swap here is the rarer kind: the model changed and the price didn't move at all.

Sources