ChatGPT vs Claude vs Gemini
Which frontier platform should anchor your stack? Scores from the RXed AI Periodic Table v1.0 — identical 20-element scale, deterministic overall. How we audit →
At a glance
Scope = elements of the AI stack in the tool’s stated scope (dashed = honestly out of scope, never a penalty). Quality = how well it scores on what it covers. No single number — both axes, always.
The verdict
Gemini posts the best pair on the table (Scope 19/20 · Quality 9/10) because Google sells the whole stack — embeddings, vector store, RAG infrastructure and multimodal all score 9+ — if you can absorb Google Cloud complexity and constant product reshuffles. ChatGPT (Scope 18/20 · Quality 8/10) is still the deepest single-vendor default: strongest fine-tuning (Ft 9) and function calling (Fc 9), with model churn as the recurring tax. Claude (Scope 19/20 · Quality 7/10) wins the agent era on the narrow-but-decisive things: the best harness (Ag 9.5, Fw 9), context engineering (Cx 9.5) and MCP — but you bring your own embeddings (Em 3) and vector store (Vx 2). Building agents → Claude. Building on one vendor's full data stack → Gemini. Wanting everything adequate in one subscription → ChatGPT.
Where the quality sits
Family mean over covered elements only — a dash means the family is honestly outside that tool’s scope.
| ChatGPT / OpenAI Platform Scope 18/20 · Quality 8/10 audited 2026-07-05 | Claude / Anthropic Scope 19/20 · Quality 7/10 audited 2026-07-07 | Gemini / Google Scope 19/20 · Quality 9/10 audited 2026-07-07 | |
|---|---|---|---|
| Vendor | OpenAI | Anthropic | Google (Alphabet) |
| Origin | US | US | US |
| Pricing | Free (with ads) · Go ~€8 · Plus ~€19 · Business ~€18–23/user · Pro ~€92–185 per month; API pay-per-token | Free · Pro ~€19 ($20/mo; $17 annual) · Max ~€95/€190 ($100/$200) · Team ~€24/€120 per seat ($25/$125) · Enterprise custom, per month; API pay-per-token | Free · AI Plus ~€5 ($4.99) · AI Pro ~€19 ($19.99) · AI Ultra ~€95/€190 ($99.99/$199.99) per month; API pay-per-token from $0.25/1M input tokens (3.1 Flash-Lite) |
| Pr Prompts | 8 | 9 | 9 |
| Em Embeddings | 8 | 3 | 9 |
| Cx Context | 9 | 10 | 9 |
| Tr Tracing | 7 | 8 | 9 |
| Lg LLM | 9 | 9 | 9 |
| Fc Function calling | 9 | 9 | 9 |
| Vx Vector store | 7 | 2 | 9 |
| Rg RAG | 8 | 7 | 9 |
| Gr Guardrails | 7 | 8 | 9 |
| Mm Multimodal | 9 | 6 | 10 |
| Ag Agents | 8 | 10 | 9 |
| Ft Fine-tuning | 9 | 3 | 9 |
| Fw Frameworks & harnesses | 7 | 9 | 9 |
| Ev Evaluations | 5 | 7 | 9 |
| Sm Small models | 8 | 8 | 9 |
| Ma Multi-agent | 7 | 8 | 9 |
| Sy Synthetic data | — | — | 6 |
| Pc Protocols | 7 | 9 | 9 |
| In Interpretability | — | 5 | — |
| Th Thinking models | 9 | 9 | 9 |
Hover any score for the written reason. Full detail: ChatGPT / OpenAI Platform passport → · Claude / Anthropic passport → · Gemini / Google passport →