Gemini 4 Argon vs Muse Wonder: Same Price, Until Nov 30

Two new models from Google and Meta went on sale a day apart. Meta announced Muse Wonder on October 12, 2026 (“Muse Wonder is available today through the Meta Model API and Muse Code”), and Google’s release notes for October 13, 2026 read “Gemini 4 Argon ( gemini-4-argon ) generally available (GA) : Released”. Their list prices are identical to the cent: $4.00 per million input tokens, $20.00 per million output tokens and $0.20 for cached input. The difference is the calendar. Google sells Argon at half that price until November 30, 2026. This post puts both models side by side with the figures as their makers print them, works out one month of the same workload on each, and places each one in its own maker’s lineup. It compares prices and published specifications only; neither maker’s pages say why the prices are what they are.

The two price lists

Per million tokens, standard paid tier, read on October 14, 2026.

Model	Input	Output	Cached input	Until	Status
Gemini 4 Argon, introductory	$2.00	$10.00	$0.10	November 30, 2026	GA
Gemini 4 Argon, list	$4.00	$20.00	$0.20	from December 1, 2026	GA
Muse Wonder	$4.00	$20.00	$0.20	no end printed	preview

Google’s block for Argon reads: “Input price | Not available | $2.00 through November 30, 2026. $4.00 starting December 1, 2026.” and “Output price (including thinking tokens) | Not available | $10.00 through November 30, 2026. $20.00 starting December 1, 2026.” The cache line is “$0.10 through November 30, 2026. $0.20 starting December 1, 2026. Implicit caching only (explicit caching is not supported).” The “Not available” column is the free tier: “Gemini 4 Argon is not available on the Free tier or on the Google AI Pro plan.”

Meta’s model page prints its table one cell per line, which reads as “muse-wonder-1.0 Available in Model API and Muse Code. 1M $4.00 $0.20 $20.00” (context, input, cached input, output). Its pricing page lists the same three figures in Standard mode and none in Contributor mode, the discounted mode Meta offers on Muse Spark: “Contributor mode isn’t available for Muse Wonder on Model API.” The page’s other row, Muse Wonder Full (muse-wonder-full-1.0), has no per-token price: Meta lists it as “Available by subscription in Muse Code.”

One month on each

The same workload on both: 100 million input tokens and 20 million output tokens in a month, standard prices. The first total uses no cache. The second assumes half of the input tokens, 50 million, are read from cache, which is a choice of this post, not a figure from either maker.

Model	Input	Output	Month, no cache	Month, half the input cached
Gemini 4 Argon, introductory	$200.00	$200.00	$400.00	$305.00
Gemini 4 Argon, list	$400.00	$400.00	$800.00	$610.00
Muse Wonder	$400.00	$400.00	$800.00	$610.00

Until November 30, the same month costs half as much on Argon: $400.00 against $800.00, or $305.00 against $610.00 with the cache. From December 1 the two rows are identical. A team that runs this workload through the whole of November pays $400.00 on Argon for that month and $800.00 on Muse Wonder.

Both caches charge 5% of the input price, but they are not switched on the same way. Google’s migration notes for Argon say “Implicit caching is enabled by default for prompts of at least 4,096 tokens and discounts cached input tokens by 95% on cache hits”, and explicit caching is not offered on this model. Meta’s pricing page says: “Cached input costs less : When part of your prompt matches a cached prefix , you pay the cached-input rate for those tokens.” Neither page promises how much of a prompt will hit the cache, which is why the second column is an assumption.

Where they differ on paper

The price is the same from December; the terms around it are not.

• Status. Argon is generally available; Google’s model page lists it as “Stable: gemini-4-argon”. Meta’s documentation says “Muse Wonder is in preview.”
• Context window. Both are 1,048,576 tokens, but Google defines its window differently from its other Gemini models: “Context window (unified input and output token limit) 1,048,576”. Input, thinking and output share it. Meta’s documentation says Muse Wonder “has a 1,048,576-token context window”.
• Maximum output. Google prints “Max output 65,536 (default synchronous) / up to 1,048,576 (with long decode )”. The Meta pages read for this post give no maximum output for Muse Wonder, and this post does not assume one.
• Long prompts. Neither charges more for a long prompt. Meta says so in words: “No long-context premium : you pay the same rate whether your context window is mostly empty or almost full.” Google’s Argon block simply prints one price per line, with no prompt-length condition.
• Free use. Argon has no free tier, as quoted above. The Meta pages read for this post print no free tier for Muse Wonder either.

Each in its own family

Each maker sells cheaper models next to its new one. Per million tokens, as read on October 14, 2026, with the same month as above and no cache:

Model	Input	Output	Cached input	Month, no cache
Gemini 3.8 Flash, promotional	$0.75	$3.75	$0.075	$150.00
Gemini 3.8 Flash, from January 1, 2027	$1.50	$7.50	$0.15	$300.00
Gemini 3.1 Pro Preview, prompts up to 200K	$2.00	$12.00	$0.20	$440.00
Muse Spark 1.3	$1.25	$4.25	$0.15	$210.00

At Google, the introductory Argon month, $400.00, costs less than the same month on Gemini 3.1 Pro Preview, $440.00, because Argon’s output is cheaper ($10.00 against $12.00) at the same $2.00 input. Gemini 3.1 Pro Preview also has a tier that Argon lacks: “$2.00, prompts <= 200k tokens $4.00, prompts > 200k tokens”. At list price, from December 1, Argon’s $800.00 is 2.67 times Gemini 3.8 Flash’s $300.00 list month, the same ratio as between the two promotional months, $400.00 and $150.00. Note that the two promotions end on different dates: Argon’s on November 30, 2026, Gemini 3.8 Flash’s on December 31, 2026. The whole Flash line, generation by generation, is in this site’s Gemini Flash post.

At Meta, the step is larger. Muse Spark 1.3’s model page reads “muse-spark-1.3 Not used to improve our products. 1M $1.25 $0.15 $4.25”, which makes this month $210.00. Muse Wonder’s $800.00 is 3.8 times that. Meta’s own guidance on choosing between them is about workload: its documentation describes Muse Wonder as for “long-running, complex tasks where capability matters more than cost per token”.

How this was read

Every price and sentence above comes from the makers’ own pages, downloaded on October 14, 2026: Google’s Gemini API pricing page, the Gemini 4 Argon model page, the “What’s new in Gemini 4 Argon” guide and the Gemini API release notes; Meta’s Muse Wonder and Muse Spark model pages, its pricing and rate limits page, its models documentation and its October 12 announcement. They match this site’s dataset rows for each model, verified on October 12 or 14. Prices are standard paid-tier list prices in US dollars, without Batch, Flex or Priority. Both models were added to this site on October 14 and appear on the new models page; the Argon increase on December 1 is recorded in advance on the price-changes page as an announced provider increase. All of Google’s current prices are on the Google pricing page of this site.

Frequently asked questions

How much does Gemini 4 Argon cost?

$2.00 per million input tokens, $10.00 per million output tokens and $0.10 for cached input through November 30, 2026, on Google's pricing page read on October 14, 2026. From December 1, 2026 it is $4.00, $20.00 and $0.20.

How much does Muse Wonder cost?

$4.00 per million input tokens, $20.00 per million output tokens and $0.20 for cached input on Meta's model and pricing pages read on October 14, 2026. Meta prints no promotion and no Contributor-mode price for it.

When does the Gemini 4 Argon promotion end?

Google prints the introductory prices 'through November 30, 2026' and the list prices 'starting December 1, 2026'. On December 1 every Argon price doubles and matches Muse Wonder's.