Model highlight

Claude Opus 5.5 for Chemical Companies: First Place, and What It Costs

Claude Opus 5.5 leads rated professional work and the priced table test, and is fifth on the general intelligence index. Its tokens cost twice Sonnet 5.5's, yet on one test the bill was the same and on another more than twice the price. For a chemical distributor or producer, the bill depends on the job.

59.1on the Artificial Analysis Intelligence Index, 5th
Anthropic · Released 22 Sep 2026

Latest update

Artificial Analysis added OpenAI's GPT-6.2 Astra at 62.1 and Google's Gemini 4 Neon at 61.4 on its Intelligence Index, so Claude Opus 5.5 moved from third to fifth at 59.1. On regulatory research vals.ai added Gemini 4 Neon at 71.21%, now the leader, so Opus is 6th of 101, inside Neon's margin, and still the dearest of the 32 results in the leading group at $20.86 a question.

1 of 12

Claude Opus 5.5 at a glance

Made by
Anthropic5
Released
22 Sep 20265
Price
$4 per million tokens sent, $20 per million written6
Reads
Text and images, up to 1 million tokens2
Settings
Five, from low to max2
Index
59.1, 5th on the Artificial Analysis Intelligence Index1
Tables
1st of 157 at the high setting3

What Claude Opus 5.5 is.

Anthropic released Claude Opus 5.5 on 22 September 2026. It reads text and images, holds up to 1 million tokens, and costs $4 for every million tokens you send and $20 for every million it writes. Anthropic says it “performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5.”

You choose how hard it thinks, from low to max. Each test on this page names the setting it ran.

Twice the token price is not twice the bill.

Opus 5.5's tokens cost twice what Sonnet 5.5's do. Three tests ran both models on the same work and published what it cost. The dashed line is what the token price predicts; the bar is what each test measured.

  1. One index task

    1x the bill

    Opus 5.5: $10.04 a task, 59.1 on the index. Claude Sonnet 5.5: $10.02 a task, 54.5 on the index.

    Artificial Analysis: models and the Intelligence IndexNo date stated

  2. One regulatory question

    1.79x the bill

    Opus 5.5: $20.86 a question, 65.15% right. Claude Sonnet 5.5: $11.67 a question, 69.70% right.

    vals.ai Legal Research, regulatory questions19 Nov 2026

  3. One document page

    2.11x the bill

    Opus 5.5: 5.79 cents a page, 76.97 on whole documents. Claude Sonnet 5.5: 2.74 cents a page, 69.21.

    ParseBench leaderboard14 Nov 2026

What the token price predicts: 2xWhat the test measured

Where Opus 5.5 leads.

Three of the guide's jobs have a test that ran it.

Table structure score, out of 100. The test's maker sells LlamaParse, which is marked

  1. Claude Opus 5.5, high setting, 6.13 cents a page94.26
  2. LlamaParse Agentic Plus, the test's maker93.27
  3. GPT-6 Astra, low setting93.18
  4. Claude Sonnet 5.591.07
  5. Gemini 3 Flash, minimal setting, 0.65 cents a page89.64

First at its high setting. A Gemini 3 Flash setting scores 4.62 points less at 0.65 cents a page, about a ninth of the price.

ParseBench leaderboard14 Nov 2026

Opus 5.5 compared with Sonnet 5.5 and Fable 5.1.

Anthropic's three largest models, on the tests that ran all three.

Claude Opus 5.5Claude Sonnet 5.5Claude Fable 5.1
Price per million tokens, in and out$4 and $20$2 and $10$10 and $50
Intelligence index59.154.549.2
Cost of one index task$10.04$10.02$12.99
Professional work rating186618381765
Tables, out of 10094.26, high setting91.0791.53
Regulatory questions right65.15%, $20.86 each69.70%, $11.67 each68.18%, $20.31 each

Where Opus 5.5 is not the pick.

First place is not always worth its price. These are the places the tests say so.

  • Regulatory Research Costs Most Here

    It scores 65.15%, inside the leader's margin, at $20.86 a question: the dearest of the 32 results in the leading group. Sonnet 5.5 scored higher for $11.67.

    vals.ai Legal Research, regulatory questions19 Nov 2026

  • Cheaper Readers Come Close on Tables

    At its high setting it leads the table test at 6.13 cents a page. A Gemini 3 Flash setting is 4.62 points behind at 0.65 cents. A free 1.2B model is first on tables in the independent document test.

    ParseBench leaderboard14 Nov 2026

  • Parsing Tools Lead Whole Documents

    On whole documents it is 7th of 157, behind six document parsing tools. Two of them are sold by the test's maker, so check a second test before you choose.

    ParseBench leaderboard14 Nov 2026

  • Long Answers Raise the Bill

    At its max setting it wrote about 119,000 tokens per index task, against about 73,000 for Opus 5. Every one of them is charged at $20 a million.

    Artificial Analysis: Claude Opus 5.5 takes the top spot on the Intelligence Index22 Sep 2026

How to test Opus 5.5 against a cheaper model.

The tests disagree on what Opus costs you, so measure it on your own work.

  1. Pick One Real Job

    Choose one job you run often: a price list to pull, a customer report, or a regulatory question.

    You end withOne job, with work your team has already checked

  2. Run It on Both

    Run 20 real examples through Opus 5.5 and through a cheaper model, at the same setting.

    You end with40 answers, side by side

  3. Count the Bill, Not the Price

    Add up what each run cost from your account's usage, not from the price list.

    You end withWhat each model really costs on your work

  4. Check Every Answer

    Mark each answer right or wrong against the work your team checked.

    You end withYour own score for each model

  5. Pay for the Gap Where It Shows

    Use Opus where its score is clearly higher, and the cheaper model where the two are level.

    You end withA model for each job, chosen on your evidence

Questions about Claude Opus 5.5.

Short answers, from the same sources as the rest of this page.

What is Claude Opus 5.5?

A general model from Anthropic, released on 22 September 2026. It reads text and images and costs $4 for every million tokens sent and $20 for every million written.

Is Claude Opus 5.5 the best AI model?

It leads rated professional work at 1866 and is fifth on the Artificial Analysis Intelligence Index at 59.1, behind Claude Fable 5.2 at 64.6, GPT-6.2 Astra at 62.1, Gemini 4 Neon at 61.4, and Gemini 4 Carbon at 60.0. It is not first everywhere: on regulatory research it scores 65.15%, behind Gemini 4 Neon at 71.21% and Gemini 4 Hydrogen and Sonnet 5.5 at 69.70%, though the test cannot separate them.

How much does Claude Opus 5.5 cost?

$4 for every million tokens sent and $20 for every million written. What a job costs depends on how much it writes: one index task cost $10.04, a regulatory question $20.86, and a document page 5.79 cents at the default setting.

Is Claude Opus 5.5 good at reading price lists?

It is first on ParseBench's table test at its high setting, at 94.26. That test's maker sells a product it ranks. On the independent document test, a free 1.2B model is first on tables.

Should I use Claude Opus 5.5 or Claude Sonnet 5.5?

On professional work they are level, and in one test an Opus task cost the same as a Sonnet task, not twice as much. On regulatory questions Sonnet scored higher for less. Test both on your own work and compare the bills.

Sources.

  1. Artificial Analysis: models and the Intelligence Index Artificial Analysis, an independent testing company. Each model's record gives its index score, its cost per index task, and its professional work rating.States no date · Vailent checked 21 Nov 2026
  2. Artificial Analysis: Claude Opus 5.5 takes the top spot on the Intelligence Index Artificial Analysis, an independent testing company, on its own tests of the model.Published 22 Sep 2026 · Vailent checked 30 Sep 2026
  3. ParseBench leaderboard LlamaIndex, which sells LlamaParse, a product it ranks.Published 14 Nov 2026 · Vailent checked 15 Nov 2026
  4. vals.ai Legal Research, regulatory questions vals.ai, an independent testing company. Lawyers wrote and checked the questions; an answer counts only if it meets every point on the lawyers' list.Published 19 Nov 2026 · Vailent checked 21 Nov 2026
  5. Anthropic: Claude Opus The maker's own page.States no date · Vailent checked 30 Sep 2026
  6. Anthropic: pricing The maker's own page.States no date · Vailent checked 30 Sep 2026