ChatGPT vs. Gemini vs. Claude: The Best LLM Subscription You Should Buy
Devansh, writing for Artificial Intelligence Made Simple, put ChatGPT, Gemini, and Claude through hundreds of real-world tasks and came away with a clear winner, a forgettable middle option, and one he considers a waste of money. For anyone trying to decide which AI subscription ...
Devansh, writing for his Substack publication Artificial Intelligence Made Simple, published a detailed breakdown on June 3, 2025 comparing the three dominant consumer AI subscriptions: OpenAI's ChatGPT, Google's Gemini, and Anthropic's Claude. Rather than leaning on benchmark scores or sponsored talking points, Devansh ran hundreds of actual tasks through each platform, including complex research, first-draft writing, and visual data requests, to determine which subscription delivers real value for real work. His verdict is blunt: one earned his upgrade to the highest tier, one is forgettable, and one he considers a straight-up waste of money.
Why This Matters
The AI subscription question is no longer hypothetical. Paying users have to choose, and choosing wrong means wasting $20 to $200 a month on a tool that underperforms for their specific workflow. ChatGPT currently holds roughly 65 percent of the active user base, a commanding lead that reflects first-mover advantage more than universal superiority. Devansh's review matters because it separates the hype from actual task performance, and his willingness to call one subscription a waste of money is the kind of honest take that most sponsored content refuses to make.
Daily briefing from 50+ sources. Free, 5-minute read.
The Full Story
Devansh structured his comparison around web app performance only, explicitly excluding API use cases. That is an important framing choice. API performance, rate limits, and developer-facing features are a completely different conversation from what a professional experiences when they open a browser tab and need something done in the next 20 minutes.
ChatGPT emerged as Devansh's top pick, compelling enough that he upgraded to its highest available subscription tier. The platform's strength lies in breadth: research queries, general knowledge, and versatile task handling across a wide range of topics. For users who need one tool that can handle most things reliably without requiring them to switch contexts, ChatGPT delivers consistently enough to justify the premium cost. OpenAI's continuous update cycle and first-mover advantage have produced a product with fewer rough edges than its competitors in general-purpose use.
Gemini, Google's entry, lands in the middle of Devansh's ranking as the forgettable option. It is not bad, but it fails to distinguish itself in ways that would make someone choose it over ChatGPT unless they are already deeply embedded in Google Workspace. Its native integration with Gmail, Google Calendar, and other Google productivity tools provides real efficiency gains for that specific audience, but for users outside that ecosystem, the case for Gemini falls apart quickly. The platform benefits from Google's infrastructure and scale, yet it has not translated those advantages into a product that stands out on raw task performance.
Claude draws Devansh's harshest verdict. Anthropic has built its reputation around safety, reduced hallucination rates, and what it calls constitutional AI development principles. Claude genuinely performs well on writing tasks, and critics of other models often point to Claude's editorial quality as its strongest suit. But Devansh's assessment for the web app subscription is that the value proposition does not hold up against what you pay. Anthropic's focus on safety and reliability over raw capability expansion has produced a product that is thoughtful but limited in ways that matter when you are trying to get actual work done across a variety of task types.
What makes Devansh's review credible is the scope of testing. Hundreds of tasks across multiple categories is not a weekend experiment. It reflects sustained, repeated use that surfaces the patterns and failure modes that a quick demo never reveals. His decision to upgrade ChatGPT to its highest tier with his own money is the clearest possible signal of where his confidence sits after that testing.
Key Details
- Devansh published this comparison on June 3, 2025, through Artificial Intelligence Made Simple on Substack.
- He tested ChatGPT, Gemini, and Claude exclusively as web applications, not through their APIs.
- ChatGPT holds approximately 65 percent of active AI user market share as of mid-2025.
- Devansh upgraded to ChatGPT's highest subscription tier based on his testing results.
- Claude's subscription was rated the weakest value proposition of the three.
- Gemini was characterized as the middle option: functional within Google's ecosystem but forgettable outside .
- Anthropic's development philosophy prioritizes constitutional AI safety principles over pure capability expansion.
What's Next
OpenAI will continue pushing ChatGPT feature updates at a pace that widens its lead over competitors for general-purpose users, making it harder for Gemini and Claude to close the gap through incremental improvements alone. Anthropic's best path forward for Claude is doubling down on specialized professional writing and editorial use cases where it genuinely outperforms, rather than competing head-to-head on breadth. Google needs to make Gemini irreplaceable for Workspace users rather than merely convenient, because convenience alone does not sustain subscription revenue when ChatGPT is one tab away.
How This Compares
Devansh's findings align with broader 2025 market data showing that no single model dominates every use case. The sophisticated user behavior emerging in mid-2025 is not brand loyalty, it is portfolio management. Power users maintain paid access to 2 or 3 models simultaneously, routing tasks to whichever tool handles them best. That is a rational response to a market where ChatGPT handles breadth, Claude handles editorial depth, and Gemini handles Google ecosystem integration.
Compare Devansh's stance to the growing school of thought that says the "best AI" question is the wrong question entirely. Reviewers like TJ Robertson, who published a model-by-model task guide in 2025, argue that users should think in terms of task categories rather than single subscriptions. Devansh partly agrees with that framing but pushes back implicitly by declaring a winner rather than endorsing the "use them all" hedge. That is a bolder and more useful take for readers who cannot afford or do not want to manage multiple paid subscriptions simultaneously.
What separates this review from most AI comparison content is the explicit rejection of benchmarks. Benchmark scores measure what models do in controlled conditions that rarely match real work. Devansh's approach of running actual research tasks, draft generation, and data visualization requests reflects how AI tools perform under the conditions that actually matter to paying subscribers. That methodology gap, real-world tasks versus benchmark tables, is where most AI comparisons fall short, and it is why his conclusion carries more weight than a leaderboard screenshot. For readers who want deeper guides on choosing the right AI platform for specific workflows, hands-on task testing like this remains the gold standard approach.
FAQ
Q: Which AI subscription is worth paying for in 2025? A: Based on Devansh's hands-on testing across hundreds of tasks, ChatGPT is the strongest general-purpose subscription and the one he upgraded to the highest tier himself. Gemini works well if you live inside Google Workspace. Claude produces quality writing but does not justify its subscription cost for most users who need broad task coverage rather than specialized editorial output.
Q: Is Claude better than ChatGPT for writing tasks? A: Claude does produce high-quality written content and is often praised for nuance and editorial style. However, as a web app subscription evaluated across a wide range of task types, Devansh found it the weakest overall value. If your entire workflow is writing and editing, Claude deserves a closer look, but for mixed task loads, ChatGPT holds up better.
Q: Do I need to pay for multiple AI subscriptions? A: Not necessarily. Sophisticated users increasingly maintain access to 2 or 3 paid models to route different tasks to the best-suited tool, but that strategy makes sense only if your workload is diverse and high-volume. For most people, one strong paid subscription, ChatGPT being the most defensible choice per Devansh's testing, covers the majority of real-world needs.
The AI subscription market is settling into a pattern where breadth beats specialization for most paying users, and ChatGPT's lead reflects that reality more than any single technical advantage. Keep an eye on AI news as Anthropic and Google push updates that could shift this calculus before the end of 2025. Subscribe to the AI Agents Daily weekly newsletter for daily updates on AI agents, tools, and automation.
Get stories like this daily
Free briefing. Curated from 50+ sources. 5-minute read every morning.
