# 0.1.4 How both AI vendors offer their models (/how-ai-works/what-ai-actually-is/both-ai-vendors) --- type: Document title: "0.1.4 How both AI vendors offer their models" description: "The model families Anthropic and OpenAI offer, each one's setting for working before an answer, and the knowledge cutoffs each states, as checked on 16 October 2026, with Anthropic's plan details checked again on 31 October 2026." status: stable order: 11.4 ksor: owner: team:panaversity audience: [ public ] approval: by: process:panaversity at: 2026-11-02T08:13:34Z chapter: "0.1" part: "0" expert_status: null # as the Part 0 overview: Part 0 is optional on every route, after the skip quiz concepts: [ "0.1.4" ] last_verified: 2026-10-16 review_by: 2026-11-16 sources: - id: anthropic-models-overview resource: https://platform.claude.com/docs/en/about-claude/models/overview title: "Models overview (Claude Platform Docs, verified 2026-10-18)" - id: anthropic-pricing resource: https://claude.com/pricing title: "Plans & pricing (Claude, verified 2026-10-31)" - id: anthropic-academy-models resource: https://academy.claude.com/tutorials/choosing-the-right-claude-model title: "Choosing the right Claude model (Claude Academy, verified 2026-10-31)" - id: anthropic-model-settings resource: https://support.claude.com/en/articles/8664678-change-the-model-effort-and-thinking-settings title: "Change the model, effort, and thinking settings (Claude Help Center, updated 2026-10-07, verified 2026-10-16)" - id: anthropic-system-prompts resource: https://platform.claude.com/docs/en/release-notes/system-prompts/claude-sonnet-5-5 title: "Claude Sonnet 5.5 system prompts, and the Claude Haiku 5.5 page beside it (Claude Platform Docs, entries of 2026-10-07, verified 2026-10-16)" - id: openai-gpt6 resource: https://openai.com/index/gpt-6-for-everyone/ title: "GPT-6 and Intelligent UI for everyone (OpenAI, 2026-10-07, verified 2026-10-16)" - id: openai-models-help resource: https://help.openai.com/en/articles/20001354-gpt-6-and-other-models-in-chatgpt title: "GPT-6 and other models in ChatGPT (OpenAI Help Center, updated 2026-10-09, verified 2026-10-16)" - id: openai-free-faq resource: https://help.openai.com/en/articles/9275245-chatgpt-free-tier-faq title: "ChatGPT Free Tier FAQ (OpenAI Help Center, updated 2026-08-13, verified 2026-10-16)" - id: openai-pricing resource: https://chatgpt.com/pricing title: "Plans and Pricing (ChatGPT, verified 2026-10-16)" - id: openai-sol resource: https://developers.openai.com/api/docs/models/gpt-6-sol title: "GPT-6 Sol model page (OpenAI API docs, verified 2026-10-16)" - id: openai-astra resource: https://developers.openai.com/api/docs/models/gpt-6-astra title: "GPT-6 Astra model page (OpenAI API docs, verified 2026-10-16)" generated: at: 2026-11-02T08:13:34Z by: process:panaversity trust_tier: unverified build_id: sha256:e1fbcb7d9a8f8b9b7cb96875022325c02c9ba4be94cef13b301cbd39a9edad85 dirty: true ksor_version: 0.0.60 --- **In everyday life.** Two coffee shops both sell small, medium and large cups, but each shop uses its own names. To compare them, you look at how much each cup holds, not at what it is called. **What you will learn.** What differs between the two AI vendors in a way that changes what you do, and what stays the same on both. A list at the end gives each AI vendor's models, free plans, thinking settings and knowledge cutoffs. **Why it matters.** When you open the model menu, you see these names. This page says what they mean on each AI vendor. The names change often, so it shows the date it was checked. Each AI vendor offers several models at once, grouped into model families, from fast and low-cost to the most capable. Each model was trained on its own, so each has its own knowledge cutoff. The AI vendor does not always state it. Most models also have a setting for how much working they write before they answer, such as an effort level or a thinking level. Choosing among the models is Chapter 2's subject. **The differences that change a decision.** Three matter in this chapter. 1. **Thinking on and off.** On Claude's current models, thinking stays on, and the effort level sets how much working is written. On ChatGPT, the Free plan has Think for harder questions, and paid plans have the Thinking slider, which starts at Instant. So you can compare an answer with and without thinking on ChatGPT, but not on Claude. 2. **Stated cutoffs.** Anthropic states one cutoff for its current models. It also puts the date in the standing instructions of Haiku 5.5 and Sonnet 5.5. So those chats can quote the date from the window. OpenAI states cutoffs for its developer models only. It does not say whether a ChatGPT window holds the date. So a cutoff that ChatGPT gives you may come from training alone. 3. **Family names.** Anthropic's families are its tiers, such as Sonnet, each released in new versions. OpenAI's families are generations, such as GPT-5.6, with tiers such as Luna and Sol inside them. So compare models by the job each AI vendor gives them, not by name. **The invariant.** The names, the settings and the dates will change. What a model works from will not change, and that is the invariant. Every model on either AI vendor works from what it learned in training, which stops at its knowledge cutoff, and from its context window. An effort setting or a thinking level changes how much working the model writes into that window, not what it can check. Anything that happened after the cutoff reaches the model only through the window. It comes from a tool, such as web search, or from what you paste in. **For reference.** *Anthropic, as verified 16 October 2026.* - **The families.** Claude comes in four families: Haiku, Sonnet, Opus and Fable. The current versions are Haiku 5.5, Sonnet 5.5, Opus 5.5 and Fable 5.1. Haiku is for fast work at high volume, and Sonnet gives the best mix of speed and intelligence. Opus is for long-running coding and knowledge work, and Fable is for demanding reasoning and the longest agent work.[^anthropic-models-overview] - **The plans.** The Free plan has Haiku and Sonnet.[^anthropic-pricing] On 31 October 2026, two Anthropic pages differed about the paid plans. The model guide says Pro and Max add Opus and Fable.[^anthropic-academy-models] The plans page lists Opus on Pro, Max and Ultra. It includes Fable on Max and Ultra, and on Pro, Fable is paid with usage credits, a pay-as-you-go balance.[^anthropic-pricing] - **Effort and thinking.** One model menu, next to the send button, sets the model, the effort level and thinking. Effort runs from Low to Max. Higher effort gives a more thorough reply, takes longer and uses more tokens, so the usage limit runs out sooner.[^anthropic-model-settings] Claude also works through its reasoning before it replies, in a "Thinking" section that you can open. In the Claude apps, this thinking stays on for Haiku 5.5, Sonnet 5.5, Opus 5.5 and Fable 5.1.[^anthropic-model-settings] Anthropic's pages do not say whether the Free plan shows the effort setting, so look at your own model menu. - **The cutoffs.** Anthropic states a reliable knowledge cutoff of June 2026 for all four current models.[^anthropic-models-overview] It also publishes the system prompts of its apps, which are their standing instructions. The system prompts for Haiku 5.5 and Sonnet 5.5 give the model the same date. They also tell it to point to web search for later events.[^anthropic-system-prompts] *OpenAI, as verified 16 October 2026.* - **The families.** OpenAI's model families are its generations, such as the GPT-5.6 family, with tiers, or levels, inside each. In that family, Luna is the fastest and lowest-cost tier. Terra balances capability, speed and cost, and Sol is for complex work.[^openai-models-help] GPT-6 Astra is OpenAI's most capable model.[^openai-astra] - **The Free plan.** OpenAI's own pages disagree about which model the Free plan runs. Its announcement[^openai-gpt6] and its pricing page[^openai-pricing] say GPT-6 Luna, and its model help page[^openai-models-help] and its Free Tier FAQ[^openai-free-faq] say GPT-5.6 Luna. So this page calls it the free plan's model. - **Thinking levels.** On paid plans, a Thinking slider in the model menu sets how much ChatGPT thinks, from Instant up to Pro. The levels you get depend on your plan.[^openai-models-help] The Free plan has no Thinking slider.[^openai-models-help] Instead, a Free user chooses Think for harder questions.[^openai-free-faq] - **The cutoffs.** OpenAI states knowledge cutoffs on its developer model pages, for example April 20, 2026 for GPT-6 Sol[^openai-sol] and April 30, 2026 for GPT-6 Astra.[^openai-astra] Its ChatGPT pages state no cutoff for the versions in ChatGPT. [^anthropic-models-overview]: Models overview, Claude Platform Docs. [^anthropic-pricing]: Plans & pricing, Claude. [^anthropic-academy-models]: Choosing the right Claude model, Claude Academy. [^anthropic-model-settings]: Change the model, effort, and thinking settings, Claude Help Center, updated 7 October 2026. [^anthropic-system-prompts]: Claude Sonnet 5.5 system prompts and Claude Haiku 5.5 system prompts, Claude Platform Docs, entries of 7 October 2026. [^openai-models-help]: GPT-6 and other models in ChatGPT, OpenAI Help Center, updated 9 October 2026. [^openai-astra]: GPT-6 Astra model page, OpenAI API docs. [^openai-gpt6]: GPT-6 and Intelligent UI for everyone, OpenAI, 7 October 2026. [^openai-pricing]: Plans and Pricing, ChatGPT. [^openai-free-faq]: ChatGPT Free Tier FAQ, OpenAI Help Center, updated 13 August 2026. [^openai-sol]: GPT-6 Sol model page, OpenAI API docs.