Four models from three makers now list $2 in, $10 out

Matching list prices; only Google’s page dates an end

In short

On 4 November Google's pricing page listed Gemini 4 Argon at $2 input and $10 output per million tokens through 30 November, and $4 and $20 from 1 December.
OpenAI's pages for GPT-6 Sol and GPT-6.1 Sol price prompts over 272,000 input tokens at twice the input rate and 1.5 times the output rate, for the full request.
Anthropic's pricing page listed Claude Sonnet 5.5 at $2 per million input tokens and $10 per million output tokens on 4 November, with no end date given for either price.

notis drafts its stories with Claude, which is made by Anthropic, one of the five companies whose price lists this article compares.

OpenAI's page for GPT-6 Sol lists $2 per million input tokens and $10 per million output tokens. 1 Its pricing page lists GPT-6.1 Sol at the same $2 and $10, under a setting for prompts of up to 272,000 input tokens. 2 Anthropic's pricing page lists Claude Sonnet 5.5 at $2 and $10. 3 Google's pricing page lists Gemini 4 Argon at $2 and $10 through 30 November, and at $4 and $20 from 1 December. 4

per million tokens · Claude Sonnet 5.5 input, 4 Nov 2026

The other three pages give no end date for their $2 and $10. 5 We opened all four on 4 November.

Above 272,000 input tokens, OpenAI's pages for both Sols price the whole request at twice the input rate and 1.5 times the output rate, which on $2 and $10 is $4 and $15. 6 Anthropic's page says its models from Claude 4.6 on, except Haiku 5.5, include the full 1 million token context window at standard pricing. 7 Google's page lists Argon's two dated standard prices with no separate price by prompt length. 8

Nor is $2 and $10 the top of OpenAI's or Anthropic's list. Anthropic lists Claude Opus 5.5 at $4 and $20 and Claude Fable 5.1 at $10 and $50, and OpenAI lists GPT-6 Astra at $10 and $50. 9

This piece carries claims checked when they were first published, with the quote each rested on and the date it was opened. Every current price in it was opened again on 4 November.

“originally $0.09 per M input tokens, now $0.038 per M input tokens”
Cloudflare

Part 1 of 530 September: Google sets Argon's price, and a rise after it

Google first announced Gemini 4 Argon on 30 September 2026, saying it was rolling out to a set of trusted cyber defenders through its Fairwind Program. 10 In that post Google said Argon would launch at an introductory price of $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 after the introductory period. 11

Our own record has one correction on this subject. A story we ran on 28 September got the baseline of an OpenAI price cut wrong in a key point, not in its headline, and we corrected it on 3 October.

A flight of weathered stone steps rising between two walls of rough-cut stone blocks, with small plants growing in the gaps, under a partly cloudy sky. — The stone steps rise between two rough walls, with small plants growing in the gaps. Photo: Tama66 / Pixabay

Part 2 of 56 to 13 October: Haiku 5.5 lists 90% below Haiku 4.5 up to 100,000 tokens, and Argon opens

Mistral AI published Mistral Large 4 on 6 October and said it would release the weights by the end of the month. 27

Anthropic published Introducing Claude Haiku 5.5 on 7 October 2026, describing it as the cheapest, fastest and most capable small model it has released. 12 Anthropic says Haiku 5.5 is priced 90% lower than Haiku 4.5 for requests up to 100,000 tokens, and 50% lower for requests over 100,000 tokens. 13

The pricing page lists Haiku 5.5 at $0.10 in and $0.50 out for prompts up to 100,000 tokens, and $0.50 and $2.50 for prompts over that. 14 It lists Haiku 4.5 at $1 in and $5 out. 15

The same 7 October page says that, starting that day, cache reads on Claude Sonnet 5.5 cost 50% less: $0.10 per million tokens rather than $0.20. 16 Anthropic's pricing page lists Sonnet 5.5's cache hits at $0.10. 17

Cloudflare's post introducing Clef-omni carries the date 9 October and lists Clef-flash as "originally $0.09 per M input tokens, now $0.038 per M input tokens". 18 Cloudflare says: "The hosted version of Clef-flash now has a context window of 24k, instead of 64k as previously advertised." 19

On 13 October Google said developers can build with Gemini 4 Argon, which it describes as its frontier model for complex workflows across software engineering, enterprise knowledge work and cybersecurity defense. 20

Part 3 of 526 to 29 October: OpenAI prices GPT-6.1 Terra at $1 and $5, and Google says Carbon will launch at $2 and $10

OpenAI published Introducing GPT-6.1 Terra on 26 October, at $1 per million input tokens and $5 per million output tokens, and says those prices are 50% lower than GPT-5.6 Terra's and GPT-6.1 Sol's. 21 OpenAI says that at max effort Terra is over three times as token efficient as Claude Opus 5.5 and one-fourth its price per token, for a roughly 9x lower cost per task. 22 That comparison is OpenAI's own; we did not run it.

On 28 October Google published "Gemini 4 Carbon: our next frontier model", which calls Carbon "our strongest model with state-of-the-art cybersecurity and long horizon capabilities". 23 Google says Carbon will launch at the same introductory price as Argon, $2 and $10, and its footnote says $4 and $20 apply after the introductory period. 24 Google says it will make Carbon available first to a subset of cyber defenders, and gives no date for developers. 25 Google's pricing page, opened on 4 November, does not list Gemini 4 Carbon. 26

Mistral's post of 29 October, releasing the weights, lists API prices of $0.70 per million input tokens, $0.07 for cached input and $2.10 for output, "starting today". 28 Mistral's model page lists the same $0.70 and $2.10. 29

Leave Google's two models aside, and each input price this piece dates to October sits below the $2 input of the four: $0.10 for Haiku 5.5, $0.038 for Clef-flash, $1 for Terra and $0.70 for Mistral Large 4.

Quoted from blog.google on 2026-10-29: After the introductory period expires, the price of $4 per 1M input… — Google's footnote on its Gemini 4 Carbon page says the $4 and $20 prices apply after the introductory period. blog.google, quoted / Gemini 4 Carbon: our next frontier model · quoted-evidence

Part 4 of 54 November: Anthropic says Haiku 5.5 counts the same text as approximately 30% more tokens than Haiku 4.5

Anthropic's documentation says Haiku 5.5 counts the same text as approximately 30 percent more tokens than Haiku 4.5. 30 Its pricing page says the exact increase depends on the content and workload shape. 31

OpenAI says reasoning tokens are billed as output tokens, Google lists Argon's output price as including thinking tokens, and Anthropic's Claude Code documentation says thinking tokens are billed as output tokens. 32

Anthropic's pricing page says a Haiku 5.5 request's prompt length counts all of its input tokens, including cache reads and writes, and that each request is priced on its own. 33

Argon's is not the only dated price on Google's page: it lists Gemini 3.8 Flash at $0.75 in and $3.75 out through 31 December, and $1.50 and $7.50 from 1 January 2027. 34

Part 5 of 5Anthropic says Haiku 5.5 costs around 75% less to run on average than Haiku 4.5

The task is ours, and we did not run it. Summarise 1,000 contracts, one request each, at 20,000 input tokens and 1,000 output tokens per request as Haiku 4.5 counts them. We assume the standard tier, no batch discount, no caching, no thinking tokens, and that Haiku 5.5 counts the same text as 30% more tokens, input and output alike.

On Haiku 4.5's $1 in and $5 out, 20 million input tokens and 1 million output tokens cost $25.00. On Haiku 5.5's $0.10 in and $0.50 out, 26 million input tokens and 1.3 million output tokens cost about $3.25. Each request is then 26,000 tokens, under Haiku 5.5's 100,000-token line. On our arithmetic, that is about an eighth of the cost ($3.25 against $25.00).

Anthropic gives an average of its own. Anthropic says Haiku 5.5 now costs around 75% less to run on average than Haiku 4.5. 35 Its footnote says the calculation accounts for Haiku 5.5's updated tokenizer, which uses slightly more tokens per task. 36

Lengthen each contract to 80,000 tokens and the request crosses the line on Haiku 5.5, at 104,000 tokens. Haiku 4.5 then costs $85.00, for 80 million input tokens at $1 and 1 million output tokens at $5. Haiku 5.5 costs $55.25, for 104 million input tokens at $0.50 and 1.3 million output tokens at $2.50. On our arithmetic, that is about two-thirds of the cost ($55.25 against $85.00).

This compares price lists, not what the task costs. Each total is our arithmetic for 20 million input tokens and 1 million output tokens.

OpenAI's pricing page lists GPT-6 Luna at $0.10 in and $0.50 out. 37 It lists GPT-6.1 Terra at $1 in and $5 out. 38

Luna at $0.10 and $0.50, and Haiku 5.5 at the same $0.10 and $0.50 for prompts up to 100,000 tokens, each come to $2.50. Mistral Large 4, at $0.70 and $2.10, comes to $16.10. Gemini 3.8 Flash, at $0.75 and $3.75 until 31 December, comes to $18.75. GPT-6.1 Terra and Haiku 4.5, both at $1 and $5, come to $25.00. The four models at $2 and $10 come to $50.00. Claude Opus 5.5 at $4 and $20, and Argon at $4 and $20 from 1 December, each come to $100.00. GPT-6 Astra and Claude Fable 5.1, both at $10 and $50, come to $250.00.

We will open Google's pricing page again on 1 December. If Argon is not listed at $4 and $20 there, the standfirst of this piece changes.

What it means for you · our view

The four models list matching prices, and the terms around them differ: Google's has an end date, OpenAI's changes above a prompt length, and Anthropic's page says its models from Claude 4.6 on, except Haiku 5.5, include the full context window at standard pricing. OpenAI's guide, Google's page and Anthropic's Claude Code documentation each bill thinking or reasoning tokens as output, so the output rate is the one to watch on reasoning work. Before you move a workload to any of the four, check its usual prompt length against OpenAI's threshold, and price Argon at its later rate if the work runs past the date Google's page gives.

Built from 7 of our storiesEvery story this piece carries a fact from, with a link to each one still public.

How we checked this · the sourcesThis piece rests on 59 checked facts from 17 sources, checked between 7 Oct 2026 and 4 Nov 2026.

Every numbered sentence above rests on the facts listed under its number. A carried fact was checked when its story was published; it says when, and we did not open its source again for this piece.

GPT-6 Sol model · OpenAI
1. 1 OpenAI's GPT-6 Sol model page lists text-token prices of $2.00 per million input tokens and $10.00 per million output tokens, opened 4 November 2026. Quote: "Price $2 • $10 Input • Output"
  checked 4 Nov 2026
  Price $2 • $10 Input • Output ... Text tokens Per 1M tokens Input $2.00 Cached input $0.20 Cache writes $2.50 Output $10.00
2. 5 OpenAI's GPT-6 Sol model page, opened 4 November 2026, gives no end date for its $2.00 input and $10.00 output prices. Quote, the whole price block: "Input $2.00 Cached input $0.20 Cache writes $2.50 Output $10.00"
  checked 4 Nov 2026
  Input $2.00 Cached input $0.20 Cache writes $2.50 Output $10.00
3. 6 OpenAI's GPT-6 Sol model page says prompts over 272,000 input tokens are priced at twice the input and cache rates and 1.5 times the output rate for the full request. Quote: "Prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request."
  checked 4 Nov 2026
  Prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request.
Pricing · OpenAI · archived copy
1. 2 OpenAI's API pricing page lists gpt-6.1-sol on the Standard tier at $2.00 per million input tokens and $10.00 per million output tokens, opened 4 November 2026. Quote: "gpt-6.1-sol | $2.00 | $10.00 | $0.10"
  checked 4 Nov 2026
  gpt-6-astra $10.00 $50.00 $1.00 $12.50 $25.00 gpt-6.1-sol $2.00 $10.00 $0.10 $2.50 $5.00
2. 2 The pricing page's context-length setting for those prices is on its short option by default, labelled for prompts of up to 272,000 input tokens. Quote: "Short: ≤272K input tokens"; the other option reads "Long: >272K input tokens".
  checked 4 Nov 2026
  aria-checked="true" class="_SegmentedControlOption_16x8u_142" aria-label="Short: ≤272K input tokens" ... aria-label="Long: >272K input tokens"
3. 5 OpenAI's pricing page, opened 4 November 2026, gives no end date for gpt-6.1-sol's Standard prices; the dated notes it does carry include a FedRAMP uplift from 1 November, GPT-5.6 Sol's promotional pricing and GPT-6 Astra for Law's promotional pricing.
  Toned down checked 4 Nov 2026
  Starting Nov 1, 2026, FedRAMP endpoints are also charged a 10% uplift. Priority processing was renamed Fast mode on July 30, 2026. GPT-5.6 Sol’s promotional pricing is available at least through November 21, 2026. ... GPT-6 Astra for Law promotional pricing is available at least through December 31, 2026.
4. 9 OpenAI's pricing page lists gpt-6-astra on the Standard tier at $10.00 per million input tokens and $50.00 per million output tokens. Quote: "gpt-6-astra | $10.00 | $50.00 | $1.00"
  checked 4 Nov 2026
  gpt-6-astra $10.00 $50.00 $1.00 $12.50 $25.00
5. 37 OpenAI's pricing page lists gpt-6-luna on the Standard tier at $0.10 per million input tokens and $0.50 per million output tokens. Quote: "gpt-6-luna | $0.10 | $0.50 | $0.01"
  checked 4 Nov 2026
  gpt-6.1-terra $1.00 $5.00 $0.05 $1.25 $2.50 gpt-6-luna $0.10 $0.50 $0.01 $0.125 $0.25
6. 38 OpenAI's API pricing page lists gpt-6.1-terra on the Standard tier, at the default short context setting, at $1.00 per million input tokens and $5.00 per million output tokens, opened 4 November 2026. Quote: "gpt-6.1-terra | $1.00 | $5.00 | $0.05"
  checked 4 Nov 2026
  Standard pane, context length Short (aria-label "Short: ≤272K input tokens", checked by default); table header "Model | Input | Output | Cached input"; row "gpt-6.1-terra | $1.00 | $5.00 | $0.05"
Pricing · Anthropic
1. 3 Anthropic's pricing page lists Claude Sonnet 5.5 at $2 per million input tokens and $10 per million output tokens, opened 4 November 2026. Quote: "Claude Sonnet 5.5 ... $2 / MTok | $10 / MTok"
  checked 4 Nov 2026
  Claude Sonnet 5.5 | $2 / MTok | $2.50 / MTok | $4 / MTok | $0.10 / MTok | $10 / MTok ... Claude 4.6 and later models (except Claude Haiku 5.5) and Claude Mythos Preview include the full 1M token context window at standard pricing.
2. 5 Anthropic's pricing page, opened 4 November 2026, gives no end date for Claude Sonnet 5.5's $2 and $10 prices. Quote, the row: "Claude Sonnet 5.5 ... $2 / MTok | $10 / MTok | $2.50 / MTok | $4 / MTok | $0.10 / MTok"
  checked 4 Nov 2026
  Claude Sonnet 5.5 The best combination of speed and intelligence $2 / MTok $10 / MTok $2.50 / MTok $4 / MTok $0.10 / MTok
3. 7 Anthropic's pricing page says Claude 4.6 and later models, except Claude Haiku 5.5, include the full 1M token context window at standard pricing. Quote: "Claude 4.6 and later models (except Claude Haiku 5.5) and Claude Mythos Preview include the full 1M token context window at standard pricing."
  checked 4 Nov 2026
  Claude 4.6 and later models (except Claude Haiku 5.5) and Claude Mythos Preview include the full 1M token context window at standard pricing.
4. 9 Anthropic's pricing page lists Claude Opus 5.5 at $4 per million input tokens and $20 per million output tokens. Quote: "Claude Opus 5.5 ... $4 / MTok | $20 / MTok"
  checked 4 Nov 2026
  Claude Opus 5.5 For long-running agentic coding and knowledge work $4 / MTok $20 / MTok
5. 9 Anthropic's pricing page lists Claude Fable 5.1 at $10 per million input tokens and $50 per million output tokens. Quote: "Claude Fable 5.1 ... $10 / MTok | $50 / MTok"
  checked 4 Nov 2026
  Claude Fable 5.1 For demanding reasoning and long-horizon agentic work $10 / MTok $50 / MTok
6. 14 Anthropic's pricing page lists Claude Haiku 5.5 at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens. Quote: "$0.10 / MTok for prompts up to 100,000 tokens | $0.50 / MTok"
  checked 4 Nov 2026
  $0.10 / MTok for prompts up to 100,000 tokens $0.50 / MTok
7. 14 The same page lists Claude Haiku 5.5 at $0.50 per million input tokens and $2.50 per million output tokens for prompts over 100,000 tokens. Quote: "$0.50 / MTok for prompts over 100,000 tokens | $2.50 / MTok"
  checked 4 Nov 2026
  $0.50 / MTok for prompts over 100,000 tokens $2.50 / MTok
8. 15 Anthropic's pricing page lists Claude Haiku 4.5 at $1 per million input tokens and $5 per million output tokens. Quote: "Claude Haiku 4.5 | $1 / MTok | $5 / MTok"
  checked 4 Nov 2026
  Claude Haiku 4.5 $1 / MTok $5 / MTok
9. 17 Anthropic's pricing page lists Claude Sonnet 5.5 prompt-cache hits at $0.10 per million tokens, opened 4 November 2026. Quote: "On Claude Opus 5.5 and Claude Sonnet 5.5, a cache hit costs 5% of the standard input price ($0.20 USD per million tokens on Claude Opus 5.5, $0.10 USD on Claude Sonnet 5.5)."
  checked 4 Nov 2026
  On Claude Opus 5.5 and Claude Sonnet 5.5, a cache hit costs 5% of the standard input price ($0.20 USD per million tokens on Claude Opus 5.5, $0.10 USD on Claude Sonnet 5.5).
10. 31 Anthropic's pricing page says the exact increase in tokens from its newer tokenizer depends on the content and workload shape. Quote: "This tokenizer produces approximately 30% more tokens for the same text. The exact increase depends on the content and workload shape."
  checked 4 Nov 2026
  This tokenizer produces approximately 30% more tokens for the same text. The exact increase depends on the content and workload shape.
11. 33 Anthropic's pricing page says a Claude Haiku 5.5 request's prompt length counts all of its input tokens, including cache reads and cache writes, and that each request is priced on its own. Quote: "A request's prompt length counts all of its input tokens, including cache reads and cache writes. Each request is priced on its own"
  checked 4 Nov 2026
  A request’s prompt length counts all of its input tokens, including cache reads and cache writes. Each request is priced on its own: a request over the threshold pays the higher prices even when part of its prompt is a cache hit
Gemini Developer API pricing · Google · archived copy
1. 4 Google's Gemini Developer API pricing page lists Gemini 4 Argon's paid-tier Standard input price at $2.00 per million tokens through 30 November 2026 and $4.00 from 1 December 2026, opened 4 November 2026; the ai.google.dev address now opens aistudio.google.com/docs/pricing. Quote: "$2.00 through November 30, 2026. $4.00 starting December 1, 2026."
  checked 4 Nov 2026
  Input price * Not available * $2.00 through November 30, 2026. $4.00 starting December 1, 2026.
2. 4 The same page lists Argon's Standard output price, including thinking tokens, at $10.00 per million tokens through 30 November 2026 and $20.00 from 1 December 2026. Quote: "$10.00 through November 30, 2026. $20.00 starting December 1, 2026."
  checked 4 Nov 2026
  Output price (including thinking tokens) * Not available * $10.00 through November 30, 2026. $20.00 starting December 1, 2026.
3. 8 Google's pricing page lists Gemini 4 Argon's Standard input and output prices with no separate price by prompt length, opened 4 November 2026. Quote: "Input price | Not available | $2.00 through November 30, 2026."
  checked 4 Nov 2026
  * Input price * Not available * $2.00 through November 30, 2026. $4.00 starting December 1, 2026. --- * Output price (including thinking tokens) * Not available * $10.00 through November 30, 2026. $20.00 starting December 1, 2026.
4. 26 Google's Gemini Developer API pricing page, opened 4 November 2026, does not list Gemini 4 Carbon: the name Carbon appears nowhere on the page, whose first model section is Gemini 4 Argon.
  Toned down checked 4 Nov 2026
  # Gemini Developer API pricing ... ## Gemini 4 Argon {% id="gemini-4-argon" %}
5. 32 Google's pricing page lists Gemini 4 Argon's output price as including thinking tokens, and says every token the model processes is billed, including thinking tokens. Quote: "Output price (including thinking tokens)" and "Every token the model processes is counted and billed, including thinking tokens"
  checked 4 Nov 2026
  Every token the model processes is counted and billed, including thinking tokens ... Output price (including thinking tokens) * Not available * $10.00 through November 30, 2026.
6. 34 Google's pricing page lists Gemini 3.8 Flash's paid-tier Standard prices at $0.75 per million input tokens and $3.75 per million output tokens through 31 December 2026, opened 4 November 2026. Quote: "$0.75 through December 31, 2026." and "$3.75 through December 31, 2026."
  checked 4 Nov 2026
  ## Gemini 3.8 Flash {% id="gemini-3.8-flash" %} ... * Input price * Free of charge * $0.75 through December 31, 2026. $1.50 starting January 1, 2027. --- * Output price (including thinking tokens) * Free of charge * $3.75 through December 31, 2026. $7.50 starting January 1, 2027.
7. 34 The same page lists Gemini 3.8 Flash at $1.50 per million input tokens and $7.50 per million output tokens from 1 January 2027. Quote: "$1.50 starting January 1, 2027." and "$7.50 starting January 1, 2027."
  checked 4 Nov 2026
  ## Gemini 3.8 Flash {% id="gemini-3.8-flash" %} ... * Input price * Free of charge * $0.75 through December 31, 2026. $1.50 starting January 1, 2027. --- * Output price (including thinking tokens) * Free of charge * $3.75 through December 31, 2026. $7.50 starting January 1, 2027.
GPT-6.1 Sol model · OpenAI
1. 6 OpenAI's GPT-6.1 Sol model page says API prompts over 272,000 input tokens are priced at twice the input and cache rates and 1.5 times the output rate for the full request. Quote: "API prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request."
  checked 4 Nov 2026
  API prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request.
Gemini 4 Argon: our next era of frontier intelligence · Google · archived copy
1. 10 Google first announced Gemini 4 Argon on 30 September 2026, saying it was rolling out to a set of trusted cyber defenders through its Fairwind Program.
  Carriedfrom Google opens Gemini 4 Argon to paid API and Ultra users · checked 14 Oct 2026, not opened again for this piece
  Today, we’re announcing our new frontier model, Gemini 4 Argon, which is rolling out to a set of trusted cyber defenders through our Fairwind Program.
2. 11 In that post Google said Argon would launch at an introductory price of $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 after the introductory period, with cached input tokens priced at 95% off the input token price.
  Carriedfrom Google opens Gemini 4 Argon to paid API and Ultra users · checked 14 Oct 2026, not opened again for this piece
  Argon will launch at an introductory price of $2 per million input tokens and $10 per million output tokens, with cached input tokens priced at 95% off input token price. ... After the introductory period expires, the price of $4 per 1M input tokens and $20 per 1M output tokens will apply.
Introducing Claude Haiku 5.5 · Anthropic · archived copy
1. 12 Anthropic published Introducing Claude Haiku 5.5 dated October 7, 2026, with the model ID claude-haiku-5-5. Quote: "October 7, 2026"
  Carriedfrom Anthropic cuts small-model prices with Claude Haiku 5.5 · checked 8 Oct 2026, not opened again for this piece
  October 7, 2026
2. 12 Anthropic describes Claude Haiku 5.5 as the cheapest, fastest and most capable small model it has released. Quote: "Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model"
  Carriedfrom Anthropic cuts small-model prices with Claude Haiku 5.5 · checked 8 Oct 2026, not opened again for this piece
  Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released.
3. 13 Anthropic says Haiku 5.5 is priced 90% lower than Haiku 4.5 for requests up to 100,000 tokens and 50% lower for requests over 100,000 tokens. Quote: "Claude Haiku 5.5 is priced 90% lower than Claude Haiku 4.5 for requests up to 100,000 tokens, and 50% lower for requests over 100,000 tokens."
  Carriedfrom Anthropic cuts small-model prices with Claude Haiku 5.5 · checked 8 Oct 2026, not opened again for this piece
  Claude Haiku 5.5 is priced 90% lower than Claude Haiku 4.5 for requests up to 100,000 tokens, and 50% lower for requests over 100,000 tokens.
4. 16 Anthropic's Introducing Claude Haiku 5.5 page, dated 7 October 2026, says that starting that day cache reads on Claude Sonnet 5.5 cost 50% less, $0.10 per million tokens rather than $0.20. Quote: "First, starting today, we're lowering the price of cache reads on Claude Sonnet 5.5. Cache reads now cost 50% less: $0.10 per million tokens rather than $0.20."
  checked 4 Nov 2026
  October 7, 2026 ... First, starting today, we’re lowering the price of cache reads on Claude Sonnet 5.5 ... Cache reads now cost 50% less: $0.10 per million tokens rather than $0.20.
5. 35 Anthropic says Haiku 5.5 now costs around 75% less to run on average than Haiku 4.5. Quote: "On average, it now costs around 75% less to run."
  Carriedfrom Anthropic cuts small-model prices with Claude Haiku 5.5 · checked 8 Oct 2026, not opened again for this piece
  Haiku 5.5 is available at a much lower price than Haiku 4.5. On average, it now costs around 75% less to run.
6. 36 A footnote on Anthropic's Haiku 5.5 page says its calculation accounts for changes in how many tokens are used to complete a given piece of work, as Haiku 5.5 has an updated tokenizer that uses slightly more tokens per task. Quote: "This calculation also accounts for changes between Haiku 4.5 and Haiku 5.5 in how many tokens are used to complete a given piece of work"
  checked 4 Nov 2026
  This calculation also accounts for changes between Haiku 4.5 and Haiku 5.5 in how many tokens are used to complete a given piece of work: Haiku 5.5 has an updated tokenizer (similar to Sonnet 5.5’s and Opus 5.5’s), which means it uses slightly more tokens per task.
Introducing Clef-omni with full multimodality, plus a faster Clef and a cheaper Clef-flash · Cloudflare · archived copy
1. 18 Cloudflare's blog post introducing Clef-omni carries the publication date 2026-10-09.
  Carriedfrom Cloudflare released Clef-omni, which takes audio and video, and cut Clef-flash to $0.038 per million input tokens. · checked 12 Oct 2026, not opened again for this piece
  "datePublished":"2026-10-09T18:27:34.129Z"
2. 18 Cloudflare lists: "Clef-flash – originally $0.09 per M input tokens, now $0.038 per M input tokens", "Clef – remains at $0.24 per M input tokens" and "Clef-omni – launched today at $0.15 per M input tokens".
  Carriedfrom Cloudflare released Clef-omni, which takes audio and video, and cut Clef-flash to $0.038 per million input tokens. · checked 12 Oct 2026, not opened again for this piece
  originally $0.09 per M input tokens, now $0.038 per M input tokens ... remains at $0.24 per M input tokens ... launched today at $0.15 per M input tokens
3. 19 Cloudflare says: "The hosted version of Clef-flash now has a context window of 24k, instead of 64k as previously advertised."
  Carriedfrom Cloudflare released Clef-omni, which takes audio and video, and cut Clef-flash to $0.038 per million input tokens. · checked 12 Oct 2026, not opened again for this piece
  The hosted version of Clef-flash now has a context window of 24k, instead of 64k as previously advertised.
Gemini 4 Argon is now available for developers · Google · archived copy
1. 20 Google said on 13 October 2026 that developers can build with Gemini 4 Argon, which it describes as its frontier model for complex workflows across real-world software engineering, enterprise knowledge work and cybersecurity defense.
  Carriedfrom Google opens Gemini 4 Argon to paid API and Ultra users · checked 14 Oct 2026, not opened again for this piece
  Starting today, developers can build with Gemini 4 Argon , our frontier model for complex workflows across real-world software engineering, enterprise knowledge work, and cybersecurity defense.
Introducing GPT-6.1 Terra · OpenAI · archived copy
1. 21 OpenAI published "Introducing GPT-6.1 Terra" on 26 October 2026.
  Carriedfrom OpenAI says GPT-6.1 Terra meets Critical cybersecurity threshold, at half GPT-5.6 Terra's price · checked 27 Oct 2026, not opened again for this piece
  October 26, 2026
2. 21 OpenAI states standard API pricing for GPT-6.1 Terra of $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens.
  Carriedfrom OpenAI says GPT-6.1 Terra meets Critical cybersecurity threshold, at half GPT-5.6 Terra's price · checked 27 Oct 2026, not opened again for this piece
  Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens.
3. 21 OpenAI says GPT-6.1 Terra's standard API prices are 50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol.
  Carriedfrom OpenAI says GPT-6.1 Terra meets Critical cybersecurity threshold, at half GPT-5.6 Terra's price · checked 27 Oct 2026, not opened again for this piece
  Its standard API prices are 50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol
4. 22 OpenAI says that at max effort GPT-6.1 Terra is over three times as token efficient as Claude Opus 5.5 and one-fourth its price per token, resulting in a roughly 9x lower cost per task.
  Carriedfrom OpenAI says GPT-6.1 Terra meets Critical cybersecurity threshold, at half GPT-5.6 Terra's price · checked 27 Oct 2026, not opened again for this piece
  At max effort, Terra is over three times as token efficient as Claude Opus 5.5, and one-fourth its price per token, resulting in a ~9x lower cost per task.
Gemini 4 Carbon: our next frontier model · Google · archived copy
1. 23 Google published "Gemini 4 Carbon: our next frontier model", by Koray Kavukcuoglu, SVP, Google DeepMind and Chief AI Architect, on 28 October 2026.
  Carriedfrom Gemini 4 Carbon opens to some cyber defenders first, with no developer date · checked 29 Oct 2026, not opened again for this piece
  Gemini 4 Carbon: our next frontier model. Oct 28, 2026. Koray Kavukcuoglu, SVP, Google DeepMind and Chief AI Architect, Google
2. 23 Google says: "Today, we’re announcing the next frontier model: Gemini 4 Carbon, our strongest model with state-of-the-art cybersecurity and long horizon capabilities."
  Carriedfrom Gemini 4 Carbon opens to some cyber defenders first, with no developer date · checked 29 Oct 2026, not opened again for this piece
  Today, we’re announcing the next frontier model: Gemini 4 Carbon, our strongest model with state-of-the-art cybersecurity and long horizon capabilities.
3. 24 Google says Gemini 4 Carbon "will launch at the same introductory price as Argon at $2 per million input tokens and $10 per million output tokens, with cached input tokens priced at 95% off input token price."
  Carriedfrom Gemini 4 Carbon opens to some cyber defenders first, with no developer date · checked 29 Oct 2026, not opened again for this piece
  Gemini 4 Carbon will launch at the same introductory price as Argon at $2 per million input tokens and $10 per million output tokens, with cached input tokens priced at 95% off input token price.
4. 24 Google's footnote says: "After the introductory period expires, the price of $4 per 1M input tokens and $20 per 1M output tokens will apply."
  Carriedfrom Gemini 4 Carbon opens to some cyber defenders first, with no developer date · checked 29 Oct 2026, not opened again for this piece
  After the introductory period expires, the price of $4 per 1M input tokens and $20 per 1M output tokens will apply.
5. 25 Google says: "Starting this week, we will make Carbon available to a subset of cyber defenders and will gradually phase access to additional trusted partners through our Fairwind Program."
  Carriedfrom Gemini 4 Carbon opens to some cyber defenders first, with no developer date · checked 29 Oct 2026, not opened again for this piece
  Starting this week, we will make Carbon available to a subset of cyber defenders and will gradually phase access to additional trusted partners through our Fairwind Program.
6. 25 Google says it is "continuing to gather feedback and iterate on guardrails before making Carbon available to developers, businesses, and consumers via our product surfaces as soon as possible", and the post gives no date.
  Carriedfrom Gemini 4 Carbon opens to some cyber defenders first, with no developer date · checked 29 Oct 2026, not opened again for this piece
  We’re continuing to gather feedback and iterate on guardrails before making Carbon available to developers, businesses, and consumers via our product surfaces as soon as possible.
Mistral Large 4 · Mistral AI · archived copy
1. 27 Mistral AI published Mistral Large 4 on 6 October 2026. Quote: "Introducing Mistral Large 4"
  Carriedfrom Mistral releases Large 4, a 1-trillion-parameter model, in public preview · checked 7 Oct 2026, not opened again for this piece
  Introducing Mistral Large 4 Back to Blog 14 min read October 6, 2026 By Mistral
2. 27 Mistral says it will release the weights by the end of the month. Quote: "We will release the weights by the end of the month."
  Carriedfrom Mistral releases Large 4, a 1-trillion-parameter model, in public preview · checked 7 Oct 2026, not opened again for this piece
  We will release the weights by the end of the month.
Mistral Large 4 weights · Mistral AI · archived copy
1. 28 Mistral's post releasing the Mistral Large 4 weights is dated 29 October 2026. Quote: "datePublished":"2026-10-29T12:00:07.000Z"
  checked 4 Nov 2026
  "datePublished":"2026-10-29T12:00:07.000Z"
2. 28 Mistral says it is releasing the weights of Mistral Large 4, which it calls its flagship 1-trillion-parameter, natively multimodal model, an open-weight hybrid instruct-and-reasoning mixture-of-experts model.
  Carriedfrom Mistral releases Large 4 weights, but paid hosting needs a separate licence · checked 30 Oct 2026, not opened again for this piece
  Earlier this month, we released Mistral Large 4 (Le Chonk), our flagship 1-trillion-parameter, natively multimodal model, in public preview. ... Today, we're releasing the weights of this improved Mistral Large 4. ... Open-weight hybrid instruct-and-reasoning MoE with multimodal input
3. 28 Mistral's post lists API prices of 0.70 US dollars per million input tokens, 0.07 dollars for cached input and 2.10 dollars per million output tokens.
  Carriedfrom Mistral releases Large 4 weights, but paid hosting needs a separate licence · checked 30 Oct 2026, not opened again for this piece
  Starting today, Mistral Large 4 is priced at $0.70 input / M, $0.07 cached input / M, and $2.10 output / M tokens.
Mistral Large 4 · Mistral AI
1. 29 Mistral's documentation page for Mistral Large 4 lists $0.7 per million input tokens, $0.07 per million cached input tokens and $2.1 per million output tokens, opened 4 November 2026. Quote: "$0.7 Input/M Tokens $0.07 Cached input/M Tokens $2.1 Output/M Tokens"
  checked 4 Nov 2026
  $0.7 Input /M Tokens $0.07 Cached input /M Tokens $2.1 Output /M Tokens
Claude Haiku 5.5 · Anthropic · archived copy
1. 30 Anthropic's Claude Haiku 5.5 documentation says the same text counts as approximately 30 percent more tokens on Haiku 5.5 than on Claude Haiku 4.5. Quote: "the same text counts as approximately 30 percent more tokens than on Claude Haiku 4.5"
  checked 4 Nov 2026
  It uses the same newer tokenizer as Claude 4.7 and later models, so the same text counts as approximately 30 percent more tokens than on Claude Haiku 4.5.
Reasoning models · OpenAI
1. 32 OpenAI's reasoning models guide says reasoning tokens are billed as output tokens. Quote: "they still occupy space in the model's context window and are billed as output tokens."
  checked 4 Nov 2026
  While reasoning tokens are not visible via the API, they still occupy space in the model’s context window and are billed as output tokens
Manage costs effectively · Anthropic · archived copy
1. 32 Anthropic's Claude Code cost documentation says thinking tokens are billed as output tokens. Quote: "Thinking tokens are billed as output tokens, and the default budget can be tens of thousands of tokens per request depending on the model."
  checked 4 Nov 2026
  Thinking tokens are billed as output tokens, and the default budget can be tens of thousands of tokens per request depending on the model.