AI has been an integral part of our lives since its inception, and to keep things going, the AI companies are also refining them so that they can better meet the requirements of the users. OpenAI, one of the first companies that brought AI into humans’ lives, has recently launched its latest update, GPT 6, which, according to the company, is ‘the world’s most intelligent and aligned model.’ With the launch, the company now stands next to Claude’s Fable 5.1 and Gemini’s 3.8 Flash. Now if you’re also wondering what sets these models apart from one another, then here is a detailed spec-based comparison of GPT 6 vs Claude Fable 5.1 vs Gemini 3.8 Flash.
An AI context window is like the AI’s short-term memory. In simple words you can say that it’s the total amount of text the AI can read, remember, and think about at one single time during your conversation. GPT 6 offers a 1,050,000 token context window, while the Gemini 3.8 Flash comes with 1,048,576 tokens, and Claude Fable 5.1 provides a 1,000,000 token context window.
Also read: Apple’s iPhone Ultra may cost over Rs 2 lakh, can come with 2TB storage and A20 Pro chip
When we talk about how much output these AI tools can generate, GPT 6 and Claude Fable 5.1 are tied at 128,000 tokens, while Gemini 3.8 Flash supports up to 65,536 tokens (about 66K). To help you understand better, let me tell you that GPT 6 and Claude Fable 5.1 can generate nearly twice as much output in a single response as Gemini 3.8 Flash.
When we talk about the knowledge cutoff GPT 6 has it up to April 30, 2026, while Claude Fable 5.1 has a cutoff of June 2026 and Gemini 3.8 Flash has a cutoff of March 2026. However, Google says that some areas may only have information up to January 2025.
OpenAI says GPT 6 hits 95.0 per cent on ARC-AGI-2, 99.9 per cent on ARC-AGI-3 and a perfect 100 per cent on ExploitBench, its cybersecurity test. On Agents’ Last Exam, it claims 59.3 per cent, against 53.6 per cent for its own older GPT-5.6 Sol.
Anthropic says Claude Fable 5.1 scores 55.8 per cent on Terminal-Bench 4.0 and 73.4 per cent on CursorBench 3.2.0, more than doubling its last version’s 24.7 per cent on the Terminal-Bench-Science test. On Humanity’s Last Exam, Anthropic reports 60.9 per cent without tools and 65 per cent with tools.
Google says Gemini 3.8 Flash reaches 90.8 per cent on Terminal-Bench 2.1, up from 81.6 per cent for the previous Flash model, and 54.9 per cent on HLE-Verified, with gains on finance and legal agent tests too.
Also read: Apple iPhone 18 Pro launch set for next week: 6 things to expect from Surprise and shine event
Users can access these models through their respective existing subscription plans. GPT-6 is available with the ChatGPT Plus plan, which starts at Rs 1,999 per month in India, while Claude Fable 5.1 is available with the Claude Pro plan, priced at Rs 2,399 per month in India. Meanwhile, Gemini 3.8 Flash is available with the Google AI Pro plan, which costs Rs 1,950 per month in India.