Research3 min read

The latest AI models compared (November 2026)

Claude Fable 5.2, Gemini 4 Carbon, GPT-6 Astra and more side by side: release dates, prices, context windows and independent scores.

By , Founder, Loopy AI

The latest AI models compared (November 2026)

Checked on 17 November 2026. Prices are each maker's standard API rates in US dollars per million tokens (input / output). Models change every few weeks, so check the makers' pages before you decide.

The short answer: Claude Fable 5.2 from Anthropic has the top independent score right now, with Google's Gemini 4 Carbon second and Claude Opus 5.5 third. For everyday work at a lower price, look at GPT-6.1 Sol, Claude Sonnet 5.5 and Gemini 4 Hydrogen.

The top models at a glance

Model

Maker

Released

Context window

API price (in / out)

Independent score

Claude Fable 5.2

Anthropic

9 Nov 2026

1M tokens

$8 / $40

65

Gemini 4 Carbon

Google

10 Nov 2026

1M tokens

$2 / $10 (rises to $4 / $20 on 1 Dec)

60

Claude Opus 5.5

Anthropic

22 Sep 2026

1M tokens

$4 / $20

59

GPT-6 Astra

OpenAI

3 Sep 2026

1.05M tokens

$10 / $50

57

Kimi K3.1 Pro

Moonshot

23 Oct 2026

1M tokens

$2 / $10

46

Muse Wonder

Meta

12 Oct 2026

1M tokens

$4 / $20

44

Qwen4-Max

Alibaba

16 Nov 2026

1M tokens

$1.50 / $4.50

43

Grok 4.7

xAI

21 Sep 2026

500K tokens

$2 / $6

35

The independent score is the Artificial Analysis Intelligence Index, which rolls many tests into one number, for each model at its best setting. Higher is better.

Everyday and budget models

Model

Maker

API price (in / out)

Independent score

GPT-6.1 Sol

OpenAI

$2 / $10

55

Claude Sonnet 5.5

Anthropic

$2 / $10

55

Gemini 4 Hydrogen

Google

$1.50 / $7.50

54

GPT-6.1 Terra

OpenAI

$1 / $5

51

Claude Haiku 5.5

Anthropic

$0.10 / $0.50

36

Muse Ember

Meta

$0.10 / $0.50

36, by Meta's figure

GPT-6 Luna

OpenAI

$0.10 / $0.50

—

Which should you use?

If you want...

Look at

The most capable model, whatever the price

Claude Fable 5.2

Top-tier results for less

Claude Opus 5.5, or Gemini 4 Carbon before its price rises on 1 December

A fast everyday model

GPT-6.1 Sol, Claude Sonnet 5.5 or Gemini 4 Hydrogen

The lowest price for high volume

Claude Haiku 5.5, Muse Ember or GPT-6 Luna

Open weights you can run yourself

Kimi K3.1 Pro

Something free in a chat app

Claude Sonnet 5.5 and Haiku 5.5 on Claude's Free plan, or Gemini 4 Hydrogen in the Gemini app

What changed in the last month

  • 23 October: Moonshot released Kimi K3.1 Pro, with its weights on Hugging Face.
  • 26 October: OpenAI released GPT-6.1 Terra.
  • 9 November: Anthropic released Claude Fable 5.2, at a lower price than Fable 5.1 ($8 / $40, down from $10 / $50).
  • 10 November: Gemini 4 Carbon became widely available.
  • 12 November: Google released Gemini 4 Hydrogen.
  • 16 November: Alibaba released Qwen4-Max in full.

Meta's top models are now called Muse, not Llama.

Coming up

  • 30 November: Claude Sonnet 4.5 is retired from Anthropic's API.
  • 1 December: Gemini 4 Argon and Gemini 4 Carbon API prices double.

How we compared them

  • Release dates, context windows and prices come from each maker's own model and pricing pages, checked on 17 November 2026.
  • The independent score is from the Artificial Analysis leaderboard on the same day. Makers' own benchmark claims aren't used in the tables.
  • We'll update this page as new models come out.

Writing code with these models? Loopy Desktop runs Claude Code, Codex and eight more coding agents side by side on one canvas.

See Loopy Desktop

Frequently asked

Which AI model is the best right now?
As of 17 November 2026, Claude Fable 5.2 has the highest score on the Artificial Analysis Intelligence Index, followed by Gemini 4 Carbon and Claude Opus 5.5.
What's the cheapest capable AI model?
Claude Haiku 5.5, Meta's Muse Ember and OpenAI's GPT-6 Luna all cost $0.10 per million input tokens and $0.50 per million output tokens through their APIs.
Which of the latest AI models can I use for free?
Claude Sonnet 5.5 and Haiku 5.5 are on Claude's Free plan, and Gemini 4 Hydrogen is in the Gemini app for free users.
Is there an open-weight model near the top?
Yes. Moonshot's Kimi K3.1 Pro has its weights on Hugging Face, and it has the highest independent score of any model from outside Anthropic, Google, OpenAI, Meta and xAI.