AI Model Comparison

Model profile & tracker

New AI Models: Everything That Shipped in November 2026

Every AI model released in November 2026: Gemini 4 Hydrogen and Neon, Qwen 4 Max, Inkling 2 Mini, and the Gestalt mystery.

Verified 2026-11-22

Updated 22 November 2026. This page is a monthly log of every notable AI model that shipped — the big closed releases, the open-weight drops and the strange leaderboard events — with prices and context windows pulled from the live comparison table. It gets rewritten as the month goes on, and archived when the next one starts.

November 2026 has been the busiest month for model launches since this site started tracking them: five notable releases in ten days, spread across three frontier labs and one open lab nobody has identified yet.

Every notable model that shipped in November 2026

Date Model Vendor Price (in/out per 1M) Context Notes
12 Nov Gemini 4 Hydrogen Google $1.50 / $7.50 1.05M GA on day one, free tier, five input modalities
16 Nov Qwen 4 Max Alibaba $1.50 / $4.50 1M 2.4T MoE, 71B active, weights closed
18 Nov Gemini 4 Neon Google $3.00 / $15.00 1M Google's frontier default, four thinking levels
18 Nov Gestalt Unknown — — Anonymous model topping the Frontier Design Benchmark
20 Nov Inkling 2 Mini Thinking Machines $0.15 / $0.50 1M Open weights (Apache 2.0), 3.4B active, runs on a laptop

One more launch is still in its promo window: GPT-6.2 Carbon's discounted rates run until 1 December, after which Google's published pricing has Neon at 25% below Carbon for equivalent work.

What matters in this batch

One of the five is open-weight. Inkling 2 Mini ships Apache-2.0 weights on Hugging Face and Ollama — the second open model in the comparison table, alongside DaisyRose 27B.

Context windows converged at a million. Hydrogen, Qwen 4 Max, Neon and Inkling 2 Mini all landed on 1M-token windows this month. A million is now the default expectation for a new model, not a differentiator.

The cheap end got cheaper. Inkling 2 Mini's $0.50 per million output tokens is 3.75 times cheaper than the next-cheapest output in the table, and it's a real open-weights model you can self-host, not a loss-leader API rate.

Nobody knows who made Gestalt. An anonymous model called Gestalt has topped Contra Labs' Frontier Design Benchmark since 18 November — 1931 Elo, a 90% win rate, 218 Elo over Claude Fable 5.2. The Gestalt page tracks the mystery.

Also this month, beyond model launches

A mathematician at Anthropic and Claude posted a claimed proof of the Riemann hypothesis on 20 November. It is unverified as of 22 November — but if the Lean development compiles, it is the largest machine-checked mathematical argument ever attempted, produced by the same class of frontier models this site tracks.

The same day, MiniMax Research published automated cryptanalysis by an internal AI model: practical breaks of seven academic cryptographic designs, including a 43,400× speedup on the McEliece challenge record. Like the Riemann claim, it awaits outside verification — and like Anthropic's cryptanalysis work before it, the model behind it is not for sale.

What to watch for the rest of the month

  • 1 December: GPT-6.2 Carbon's promo pricing ends. Neon's published rates undercut it from that date.
  • Gestalt's maker: whoever is behind the Frontier Design Benchmark leader tends to reveal within weeks.
  • Inkling 2 Small: Thinking Machines has a mid-size sibling between Mini and the full Inkling 2 pending.

FAQ

What is the newest AI model? As of 22 November 2026 it is Inkling 2 Mini, released 20 November by Thinking Machines Lab. The newest frontier model is Gemini 4 Neon, released 18 November.

What new AI models came out in November 2026? Gemini 4 Hydrogen (12 Nov), Qwen 4 Max (16 Nov), Gemini 4 Neon (18 Nov), Gestalt (18 Nov) and Inkling 2 Mini (20 Nov). The table above has prices and context windows for each.

What is the cheapest new AI model? Inkling 2 Mini at $0.15 / $0.50 per million tokens — and its weights are open under Apache 2.0.

When did Gemini 4 come out? The family shipped in stages: Hydrogen on 12 November, Neon on 18 November 2026. The Neon page has the full spec sheet.

Where can I compare them all? The home page comparison chart has all ten models in the current rotation, sorted by price, with context windows and benchmark notes.