From Budget Bots to Premium Intelligence: What the Price Hike Means
The end of cheap AI describes a market shift in which leading providers stop passing efficiency savings to customers and instead raise prices on previously low-cost models while packaging “smarter” capabilities as premium products that demand clear business returns. Google’s Gemini pricing increase is a clear sign of that shift. Gemini 3.5 Flash, the model that is supposed to power high-volume, low-stakes tasks, now costs three times more per input and output token than the Gemini 3 Flash model it replaced. At the same time, subscription tiers like AI Plus at USD 8 (approx. RM37) per month and AI Pro at USD 20 (approx. RM92) position higher usage limits and model access as paid upgrades rather than default entitlements. For businesses, the question is changing from “Is AI affordable?” to “Does each extra tier of capability pay for itself?”.
Gemini 3.5 Flash: When the ‘Cheap’ Model Isn’t Cheap Anymore
Gemini 3.5 Flash’s rate card shows how far AI subscription cost has moved. It is priced at USD 1.50 (approx. RM7) per million input tokens and USD 9 (approx. RM41) per million output tokens, compared with USD 0.50 (approx. RM2) and USD 3 (approx. RM14) for Gemini 3 Flash. That is a threefold jump at the tier designed for bulk work like summarisation, tagging, and classification. According to Artificial Analysis, running its benchmark suite on Gemini 3.5 Flash cost about USD 1,550 (approx. RM7,100), while the higher-tier Gemini 3.1 Pro came in at around USD 890 (approx. RM4,100). Because modern AI agents loop, call tools, and generate more tokens, higher per-token rates compound across each step. The result: the model branded as the budget option can become the most expensive to operate in real agentic workloads.
Real Enterprise Gemini Value: 1.4 Million Tasks Proving ROI
While pricing climbs, production data shows why many organisations still see strong enterprise Gemini value. Workflow platform Zenphi reports that AI agents running inside its system now complete 1.4 million business tasks every month in live environments. These are operational activities across healthcare, education, logistics, technology, and professional services, including document extraction, classification, summarisation, proposal drafting, decision support, and data processing. Crucially, the agents sit inside governed workflows: AI handles defined processing steps, with clear inputs, success criteria, human checkpoints, and audit trails. This architecture makes costs more predictable and outcomes more reliable, which helps justify higher Gemini pricing. Instead of speculative experiments, businesses are buying repeatable task automation at scale. When millions of tasks run every month, even modest time or error reductions can outweigh rising AI subscription cost—if usage is disciplined and measured.
Gemini Subscriptions Go Premium: Bundles, Limits, and Trade-offs
On the consumer and prosumer side, Google has shifted Gemini into a premium bundle strategy. Paid plans sit above the free Gemini tier and gate higher limits and model access. AI Plus at USD 8 (approx. RM37) per month roughly doubles usage limits versus the free plan, while AI Pro at USD 20 (approx. RM92) offers around four times the limits and higher priority access to models like Gemini 3.5 and 3.1. At the top are AI Ultra subscriptions at USD 100 (approx. RM460) and USD 200 (approx. RM920) per month, aimed at developers, creators, and technical professionals who need maximum throughput and advanced features. To soften higher fees, Google has begun bundling perks such as YouTube Premium and other services into certain Gemini tiers, framing them as lifestyle and productivity packages rather than bare AI access, and reinforcing the idea that the era of cheap AI is over.
How Businesses Should Decide: Paying More Only When ROI is Proven
With Gemini pricing increase trends and similar moves from other labs, businesses need a stricter business AI ROI playbook. First, map AI use cases to model tiers: reserve expensive models like Gemini 3.5 Flash or 3.1 Pro for tasks where quality, reasoning, or long context clearly improves revenue, risk control, or customer experience. Second, measure cost per completed workflow, not per token. The Zenphi example shows that when AI is one governed step in a process, you can compare its cost directly against human time and error rates. Third, avoid overbuying subscriptions: AI Plus or AI Pro may be enough for most teams, while Ultra plans make sense only when throughput or priority access is mission-critical. In this new market, the winning strategy is to treat AI tiers like any other enterprise software: upgrade only when data proves the return.