Google released four Gemini 4 models in five weeks: Argon on October 13, Carbon on November 10, Hydrogen on November 12 and Neon on November 18. They share one API shape, a 1M-token window and the same tools. They differ in how much they think per turn, how fast they answer and what they cost.
On Artificial Analysis's measurements, the family now sorts itself:
* Hydrogen for everyday and high-volume work. Its default high scores 50.8 at $1.87 per task, and it's the fastest per token.
* Neo
Google names Gemini 4 Neon as the upgrade target for Gemini 3.1 Pro, for general coding work on Gemini 4 Argon, and for complex coding or science work on Gemini 3.8 Flash. Changing the model string to gemini-4-neon is the easy part. The rest falls into four groups.
* Silent: the default thinking level is now high. Argon and 3.8 Flash defaulted to medium. Unless you set a level, each request thinks more than it would at medium. For 3.8 Flash users that means higher bills; for Argon users the p
Gemini 4 Neon isn't a new model line. Google's model card says so in one line: "Gemini 4 Neon is adapted from Gemini 4 Carbon." On Artificial Analysis, Neon at its default high setting scores 59.6 on the Intelligence Index against 60.0 for Carbon High, and finishes an average task in 14.8 minutes against 36.6.
The price comparison depends on the date. Today Neon costs 50% more per token than Carbon, which is still on introductory rates. On December 1 Carbon's rates double, and Neon's $3/$15 be
Sol costs one-quarter of Astra per token. Compare repository coding, security and visual-task evidence to decide when the flagship premium is justified.
Qwen4-Max has the lower ordinary token price. Kimi K3.1 Pro has a different thinking contract and more published capability metadata on AIHubMix. Neither fact tells you which produces an acceptable patch in your repository. Choose by the cost of a change that passes review, then check that your agent's API path preserves the model's required history.
The useful comparison is a concrete task: repair a bug across two files, retain public behavior, and pass the existing tests. Give both models th
Moving from Qwen3.8-Max to Qwen4-Max changes more than the model ID. Both have a million-token context, but an output limit previously sized for the answer now also pays for thinking. The default effort changes, and agents must preserve reasoning history correctly across tool calls.
Start with one Chat Completions request, then one small repository task. Check the answer, stopping reason, history, and resulting diff before moving an existing workflow.
Recalculate the output limit
The Chat
Qwen4-Max is worth evaluating if you already use Qwen3.8-Max: ordinary input and output prices are 25% lower at Alibaba Cloud's Singapore list rates. The context window stays at one million tokens, and both accept text, images, and video. That makes price the clearest documented reason to try the upgrade. It does not establish that the new model solves your tasks better or uses fewer tokens.
The migration also changes an integration contract. An answer-only output cap can now limit thinking an