Zest

CLAUDE · PRICING

Claude Opus pricing
Opus 5.5 and older.

Claude Opus 5.5 costs $4 per million input tokens and $20 per million output tokens.

Start free
npx @winding-labs/zest

Free for 3 seats · No credit card · One command signs in every agent on your machine

Sample data

Spend in the Zest app: your agent usage at list prices, by week and by model

USED BY ENGINEERING TEAMS AT

Flyps Charm Wildfires Geoforge MisaLabs

How much does Claude Opus cost per million tokens?

Claude Opus 5.5 costs $4 per million input tokens and $20 per million output tokens. A five-minute cache write costs $5, a one-hour write $8 and a cache read $0.20; the Batch API halves each, and fast mode doubles input and output for up to 2.5x faster output.

Claude Opus 5.5 API prices per million tokens, from platform.claude.com. Checked November 21, 2026.
PriceStandardBatch APIFast mode
Input$4$2$8
Output$20$10$40
Cache write, 5 minutes$5$2.501.25x fast input
Cache write, 1 hour$8$42x fast input
Cache read$0.20$0.100.05x fast input

What does a Claude Opus request cost?

It depends on how much of the prompt is cached. This calculator opens on Opus 5.5 with our own month's average Claude Code request; change the tokens to price yours:

WORK IT OUT

It opens on our own month's average Claude Code request: 177,130 cached input tokens, 3,486 written to the cache and 878 out.

On Opus 5.5, one request costs $0.0704, and 1,000 a month cost $70.42.

The same request on every current Claude model at list price.
ModelPer requestPer month
Fable 5.2$0.1054$105.42
Opus 5.5$0.0704$70.42
Sonnet 5.5$0.0352$35.21
Haiku 5.5$0.0132$13.23

What does a customer-facing agent cost on Opus?

About $1,440 a month for 10,000 six-turn conversations with a cached system prompt, against $3,240 without the cache; the cache write and the output are 68% of it. Our example, worked from Anthropic's prices:

A support agent on Opus 5.5, priced from the table above (5-minute cache)
10,000 conversations a month, 6 turns each, a 10,000-token system prompt
Every turn: 1,500 new input × $4 + 400 output × $20 per million = $0.014
Turn 1 writes the prompt to the cache: 10,000 × $5 per million = $0.050
Turns 2 to 6 read it: 10,000 × $0.20 per million = $0.002 each
One conversation: 6 × $0.014 + $0.050 + 5 × $0.002 = $0.144
Month: 10,000 × $0.144 = $1,440.00
Without the cache, every turn pays $4 per million for the prompt: $3,240.00

How has Claude Opus pricing changed?

Opus got cheaper twice: from $15 and $75 on Opus 4 and 4.1, to $5 and $25 from Opus 4.5 through Opus 5, to $4 and $20 on Opus 5.5. Our own price checks first saw Opus 5.5 at today's price on October 5, 2026, and it hasn't changed since.

Claude Opus API prices per million tokens by model, from platform.claude.com (older models under its Additional models). Checked November 21, 2026.
ModelInputOutputCache readFast mode (input / output)
Opus 5.5$4$20$0.20$8 / $40
Opus 5$5$25$0.50$10 / $50
Opus 4.8$5$25$0.50$10 / $50 (deprecated)
Opus 4.5, 4.6 and 4.7$5$25$0.50Not offered
Opus 4 and 4.1$15$75$1.50Not offered

How does Opus compare with Fable, Sonnet and Haiku?

Opus 5.5 costs half of Fable 5.2, twice Sonnet 5.5 and forty times Haiku 5.5 on short prompts, per token. Anthropic positions Opus for long-running agentic coding and knowledge work, Fable for demanding reasoning and long-horizon agentic work, Sonnet as the best combination of speed and intelligence and Haiku for high-volume, latency-sensitive tasks.

Claude's four current models per million tokens, from platform.claude.com. Checked November 21, 2026.
ModelInputOutputAgainst Opus 5.5
Fable 5.2$8$402x
Opus 5.5$4$201x
Sonnet 5.5$2$100.5x
Haiku 5.5$0.10$0.500.025x

Does long context cost more on Opus?

No: Opus 5.5 bills its full 1M-token context window at standard rates, as every model from Claude 4.6 on does except Haiku 5.5. Fast mode's premium applies across the whole window too.

What does Opus cost in Claude Code?

On an API key, every Opus token at the prices above; on Pro, Max or Ultra, nothing beyond the plan until you pass its limits. Opus 5.5 was 99.1% of the tokens in our own month: Our own month: 2,228 Claude Code sessions on the Mac that runs our coding agents, October 22 to November 20, 2026, each response counted once and priced at Anthropic's list prices. Every session ran signed in to a Claude plan, so this is what the same work would cost on an API key, not our bill. It is a heavy, agent-run month: 72% of responses came from subagents.

Our Claude Code month, October 22 to November 20, 2026: 2,228 sessions on the Mac that runs our coding agents, priced at Anthropic's API list prices. List price, not our bill. Checked November 21, 2026.
MeasuredValue
Sessions and API responses2,228 sessions, 874,770 responses
Tokens158.8 billion, 97.6% of them cache reads
ModelOpus 5.5 for 99.1% of tokens
Cost at API list prices$63,936
The same tokens with no caching$647,366
Where the cost wentCache reads 48%, output 24%, cache writes 28%
Median session$7.25 (59% of sessions under $10)
Median day$1,806
Days with a usage-limit notice16 of 30

Can you use Opus through a subscription instead of the API?

Yes: Pro, Max, Ultra, Team and Enterprise all include Opus 5.5, inside the plan's limits; Free doesn't. Past the limits, usage credits bill Opus at the API prices above.

Claude's plans per person or seat, US prices before tax, from claude.com/pricing. Checked November 21, 2026.
PlanPriceUsageClaude Code
Free$0Limited, every five hours and every weekNo
Pro$20 a month, or $17 a month billed yearly ($200 up front)Pro's allowance every five hours and every weekYes
Max$100 a month, monthly only5x Pro every five hours and every week, plus $100 a month in API creditsYes
Ultra$200 a month, monthly only10x Pro every week, no five-hour limit once verified, plus $200 a month in API creditsYes
Team Standard$25 a seat a month, or $20 billed yearly (2 to 150 people)More than Pro, plus pooled API creditsYes
Team Premium$125 a seat a month, or $100 billed yearly5x a Standard seatYes
Enterprise$20 a seat a month, billed yearly, 20 seats minimumNone included: every token at API ratesYes

Is the API or a subscription cheaper for Opus-heavy work?

A subscription, for anyone working daily: at Anthropic's average Claude Code day of $13, a month on the API passes Max's $100 in 8 days and Ultra's $200 in 16. The API wins for apps and batch jobs, where Batch halves the price. Three teams, worked out:

Three Claude Code teams at Anthropic's average of $13 per developer per active day (code.claude.com/docs/en/costs), 20 active days a month, against the plans on claude.com/pricing. An estimate, not a quote. Checked November 21, 2026.
TeamAt API ratesOn plans
1 developer$260 a monthMax $100 or Ultra $200
10 developers$2,600 a month10 Team Premium seats: $1,000 a month billed yearly
20 developers on Enterprise$5,200 a month$400 a month in seats, plus the same usage at API rates

What does Opus cost on top of tokens?

Web search costs $10 per 1,000 searches, US-only inference 1.1x every token, and Managed Agents sessions $0.08 an hour of runtime; code execution is free alongside web search or web fetch. Tools you run yourself cost only tokens: their definitions, calls and results bill as input and output, plus a fixed tool-use system prompt on each request.

What Anthropic's API bills on top of tokens, from platform.claude.com's pricing page. Checked November 21, 2026.
ExtraPrice
Batch APIHalf price on input and output
Fast mode$8 / $40 on Opus 5.5, $16 / $80 on Fable 5.2 (2x standard)
US-only inference (inference_geo: "us")1.1x every token price, on 4.6 and later models
Web search$10 per 1,000 searches, plus the tokens it returns
Code executionFree with web search or web fetch; otherwise 1,550 free hours a month, then $0.05 an hour
Claude Managed AgentsTokens at these rates, plus $0.08 per session-hour running
The 1M-token context windowNo extra charge on 4.6 and later models, except Haiku 5.5 over 100K tokens

Why is Claude Opus expensive, and is it worth it?

Opus costs more per token because it is Anthropic's larger model for hard, long-running work; whether that is worth it depends on how much of your work needs it. For routine edits, Sonnet 5.5 at half the price is often enough; Anthropic positions Opus for long-running agentic coding, and it carried nearly all of our own month.

Is Claude Opus cheaper than GPT?

No: OpenAI's GPT-6.1 Sol, the model at Opus's tier and the one Codex's Plus plan names, costs half of Opus 5.5 per token, while the largest models cost the same. Google lists Gemini 4 Carbon, its most capable reasoning model, at $2 and $10 per million tokens through November 30, 2026, half of Opus 5.5, and at Opus 5.5's $4 and $20 from December 1. Different tokenizers mean the same prompt counts differently on each.

Claude and OpenAI API prices per million tokens (input / output), standard tier. Claude's from platform.claude.com, checked November 21, 2026; OpenAI's from developers.openai.com. Checked November 18, 2026.
TierClaudeOpenAI
LargestFable 5.2: $8 / $40GPT-6.2 Astra: $8 / $40
LargeOpus 5.5: $4 / $20GPT-6.1 Sol: $2 / $10
MidSonnet 5.5: $2 / $10GPT-6.1 Terra: $1 / $5
SmallestHaiku 5.5: $0.10 / $0.50GPT-6 Luna: $0.10 / $0.50

Does Opus cost the same on Bedrock and Vertex AI?

Not always: Amazon Bedrock and Google Cloud's own Claude offering set their own prices, with a 10% premium on regional and multi-region endpoints. Fast mode for Opus 5.5 is available on the Claude API, Claude Models on Google Cloud run by Anthropic, and Microsoft Foundry on Azure.

What do users complain about with Opus on a plan?

The limits, more than the price: how fast you reach a limit depends on the model, so heavy Opus users hit the five-hour window and the weekly cap, and Max never feels unlimited. Ultra removes the five-hour limit once verified; past the week, usage credits bill Opus at API rates. Claude usage limits explains each window.

How do you see what Opus costs per engineer?

Opus dominates most Claude Code bills, so a cost per engineer is mostly an Opus number.

Zest prices every Claude Code session it captures at list rates, inside a plan's allowance too, per engineer, and ties each session to the pull request it led to. Where Claude Code sends the limit meters, on Pro and Max, Zest keeps them with each session, so a lead sees who keeps running out of the five-hour window and sizes each seat to the person.

TRY IT ON YOUR TEAM

See what your team ships with AI.

Starter is free (3 seats, no credit card). Pro is $49 a month with 5 seats and Max is $149 a month with 10 seats — priced per workspace, not per engineer, with extra seats at $19 a month. See pricing

Start free
npx @winding-labs/zest

A FEW THINGS TO KNOW

Questions

Can you use Claude Opus for free?+

No. Free runs Sonnet and Haiku; Opus needs Pro or above, or the API (new API users get free test credits).

Do Max or Team have yearly discounts?+

Team does: Standard $20 a seat and Premium $100 billed yearly. Max and Ultra bill monthly only.

What does Enterprise add over Team Premium?+

SCIM, audit logs, role-based access, spend limits per user and custom data retention, at $20 a seat billed yearly plus every token at API rates.

Do paid Claude plans include Opus on the API?+

Max, Ultra and Team include monthly API credits ($100, $200 and up to $500 pooled) that pay for Opus on the API; Pro includes none.

Can you use Claude Opus in GitHub Copilot?+

Yes: Copilot's paid plans list Opus 5.5 in the model picker, and it draws on Copilot's AI credits at $0.01 a credit, not on a Claude plan or an Anthropic API key. Copilot Pro is $10 a month with 1,000 base AI credits, and Pro+ $39 with 3,900.

When did Opus 5.5's price take effect?+

Our first check that listed it, on October 5, 2026, already showed $4 and $20; it hasn't changed since.