Last refreshed: November 5, 2026 (Pacific) — this refresh updates the $200 seat tier (now Ultra, not Max 20x) and moves the monthly API credits from announced to live.
Anthropic launched Claude Haiku 5.5 on October 7, 2026. Seat prices on claude.com did not move. The operator event is a new SKU at a lower unit price, plus a cache-read cut on Sonnet 5.5.
Haiku 5.5 is a new SKU. It is not a rename of Haiku 4.5. Haiku 4.5 remains on the list at $1 / $5. Anthropic says about 90% of prior Haiku requests were under 100,000 tokens, which is the cheaper band.
Haiku 4.5
Haiku 5.5 ≤100k
Haiku 5.5 >100k
Input / MTok
$1
$0.10
$0.50
Output / MTok
$5
$0.50
$2.50
Cache write (5-minute)
$1.25
$0.125
$0.625
Cache read
$0.10
$0.01
$0.05
Batch input / output
$0.50 / $2.50
$0.05 / $0.25
$0.25 / $1.25
Sonnet 5.5 input and output did not move. Cache reads did: $0.20 to $0.10 per million tokens. Anthropic says that is about 20% cheaper on most agentic work because cache reads are a large share of the bill. Sonnet 5 cache reads stay $0.20. Opus 5.5 stays $4 / $20 with cache reads at $0.20.
Do not average the two Haiku 5.5 bands. A prompt over 100,000 tokens pays the higher line. Haiku 5.5 also uses the newer tokenizer, so the same text can be more tokens than Haiku 4.5. The posted rate is not the whole bill.
What did not change
Free $0. Pro $20 monthly / $17 annual. Max $100 (5x Pro) and Ultra $200 (10x Pro weekly). Team Standard $20 annual / $25 monthly per seat. Team Premium $100 annual / $125 monthly. Enterprise $20/seat plus usage at API rates. Cursor Pro stays $20; Pro+ $60; Ultra $200; Teams $40/user on cursor.com/pricing, with Teams Premium $120 on the docs page.
The monthly Claude Platform credits announced at launch are live: Max $100/month, Ultra $200/month, and Team seats pool $20 per Standard seat and $100 per Premium seat, capped at $500/month. Claim them by linking a Claude Console organization in billing settings; unused credits expire each cycle, with no rollover. They are a credit, not a seat-price cut — terms in the Help Center article.
Operator note
If a loop is pinned to Haiku 4.5, it is still on the $1 / $5 meter. Move it to Haiku 5.5 only if the workload can take the new model and you watch the 100k prompt line. Cursor on-demand that bills Anthropic list will pick up the new SKU when Cursor exposes it. Cursor seat prices did not move with this launch.
A low-cost AI subscription can produce text. It cannot settle a minutes policy, absorb a week of review labor, or deliver a public record that is ready to publish.
On September 29, the Farmersville City Council voted 4–1 to pay $30 a month — $360 a year — for Grok to draft verbatim meeting minutes from council recordings. City Clerk Rochelle Giovanni would review each draft. Interim City Manager Kevin Northcraft called the AI-generated minutes “more objective” than minutes written by a person. Elon Musk quote-posted the news with one word: “Grok.” [1]
The reaction was predictable. Some people treated the vote as proof that artificial intelligence had made a routine government task nearly free. Others treated it as a punch line. Both reactions missed the operational story.
The $30 vote was never really about drafting
Farmersville had spent months arguing about what its minutes should contain. The city used summary minutes, but councilmembers pressed to include particular actions and verbatim statements. Under that pressure, work that had taken the clerk about two hours stretched to nearly a week.
That is the number that matters. The city did not merely have a transcription problem. It had a governance problem: disagreement over the purpose and level of detail of the official record. A language model can turn audio into text. It cannot decide which minutes policy the council should adopt, apply that policy with institutional judgment, or end a political dispute over whose words deserve inclusion.
The subscription buys a draft, not minutes
In the city’s comparison, Grok produced eight pages, Otter produced nine, a service identified as Government Clerk produced 12, and a human Rev.com transcript ran 111 pages. Those page counts compare outputs, but they do not answer the central question: What kind of record is the city trying to produce?
The California City Clerks Association’s minutes guidance is clear. The primary purpose of minutes is to memorialize decisions. It identifies action minutes and brief summary minutes as appropriate styles and says verbatim minutes should not be used. Action minutes record final decisions; brief summary minutes preserve the main points that led to a decision without turning the record into a transcript. [2]
A transcript records what was said. Minutes record what the body did.
That distinction matters because more text is not automatically more transparency. An audio or video recording can preserve the full discussion. Minutes serve a different function: they create a concise, reliable record of attendance, items considered, actions taken, votes, and follow-up. If a city wants verbatim minutes anyway, it should make that policy choice openly — and price the work required to review them.
The hidden cost is the clerk’s review time
Farmersville’s clerk still reviews every Grok draft. She must catch misheard names, incorrect motions, missing vote details, speaker confusion, and language that does not match the city’s adopted style. She also has to format the document, reconcile it with the agenda and staff materials, route it for approval, and prepare it for publication and retention.
None of that appears in the $30 subscription price. The software cost is visible because it arrives as a line item. The clerk’s time disappears inside payroll, backlog, and delayed work elsewhere. A city can save money on the first pass while still spending a week producing the finished record.
This is why “AI wrote the minutes” is the wrong test. The useful question is whether the city received accurate, policy-compliant minutes that a clerk can stand behind without rebuilding the document.
The useful product is a finished public record
For a small city, the product should not be a block of generated text. It should be a complete workflow: the meeting recording is transcribed; a draft is shaped to the city’s adopted minutes standard; motions, votes, names, and action items are checked; the document is formatted to the city’s template; a human verifies the result; and a publish-ready file is delivered.
The artificial intelligence does not need to be perfect. The system needs to be designed so a qualified human catches what it misses. That is not a retreat from automation. It is the control that makes automation usable in public records work.
A finished-minutes service also makes the cost legible. Instead of buying a subscription and hoping staff time falls, the city buys an outcome: reviewed minutes in the required format, ready for the clerk’s final approval and the next agenda packet.
Farmersville opened a door for hundreds of small cities
California has 482 incorporated cities. Many operate with small administrative teams, tight meeting cycles, and clerks whose responsibilities extend far beyond minutes. Farmersville’s dispute is unusually visible, but the underlying problem is common: the recording is easy to make; the official record still takes judgment and time. [3]
The 4–1 vote proved that a budget line for AI-assisted minutes can survive a public meeting. The $30 subscription opened the door. The larger opportunity sits behind it: the nearly weeklong burden that nobody priced.
The next step is to buy the outcome
Council minutes as a finished, human-verified service exist now. For a city spending days turning recordings into disputed drafts, the useful conversation is not which model can generate the most pages. It is what a publish-ready minutes workflow would remove from the clerk’s desk — and what standard the council wants that workflow to follow.
Last refreshed: November 13, 2026 — applies Fable 5.2 flagship pricing ($8/$40), verified Ultra tier ($200/mo), and current 1M/128K lineup alongside Sonnet 5.5.
Direct Answer (November 13, 2026):Claude Sonnet 5.5 (claude-sonnet-5-5) lists at $2.00 input / $10.00 output per MTok with a 1M context window and 128K max output, $0.10/MTok cache reads (5% of base input), $2.50/$4.00 cache writes, and $1/$5 batch pricing (API pricing).
TAKEAWAY PDF & MARKDOWN TWIN
Claude Sonnet 5.5 Lists at $2 / $10 — Takeaway PDF
Anthropic launched Claude Sonnet 5.5 on September 28, 2026. Seat prices on claude.com did not move. The operator event is a new SKU, not a unit-price cut: Sonnet 5.5 lists at $2 per million input tokens and $10 per million output tokens — the same pair as Sonnet 5.
Sonnet 5.5 is a new SKU, not a rename of Sonnet 5. Anthropic says it runs 30%+ faster than Sonnet 5 and costs up to 30% less per task because it uses fewer tokens, not because the published rate fell.
Sonnet 5
Sonnet 5.5
Input / MTok
$2
$2
Output / MTok
$10
$10
5-minute cache write
$2.50
$2.50
1-hour cache write
$4
$4
Cache read
$0.20
$0.10
Batch
$1 / $5
$1 / $5
One list price did move after launch: cache reads on Sonnet 5.5 were cut from $0.20 to $0.10 (0.05x base input) with the October 7, 2026 rate card — two days after this brief first published. Do not treat “30% less per task” as a 30% list-price cut. The meter is still $2 / $10. US-only inference stays 1.1x on Claude 4.6 and later models.
What did not change
Free $0. Pro $20 monthly / $17 annual. Max $100 (5x Pro). Ultra $200 (10x Pro weekly). Team Standard $20 annual / $25 monthly per seat. Team Premium $100 annual / $125 monthly. Enterprise $20/seat plus usage at API rates. Current sibling models (all 1M context / 128K output): Fable 5.2 lists at $8 / $40 ($0.20 cache reads, superseding legacy Fable 5.1 at $10 / $50 and $0.25 cache reads), Opus 5.5 lists at $4 / $20 ($0.20 cache reads), and Haiku 5.5 lists at $0.10 / $0.50 for prompts up to 100k tokens ($0.50 / $2.50 above 100k); legacy Haiku 4.5 remains $1 / $5 (200K context). Cursor Individual stays $20; Cursor Teams stays $40/user on cursor.com/pricing.
Operator note
If you are pinned to claude-sonnet-5, you are still on the $2 / $10 meter. Switch to claude-sonnet-5-5 if the workload can move and you want the newer model. Cursor on-demand that routes to Anthropic list will pick up the new SKU when Cursor exposes it — Cursor seat prices did not move with this launch.
Last refreshed: November 13, 2026 (Pacific) — updates sibling flagship reference to Fable 5.2 ($8/$40), verified Ultra tier ($200/mo), and current 1M/128K lineup alongside Opus 5.5.
Direct Answer (November 13, 2026):Claude Opus 5.5 (claude-opus-5-5) is Anthropic’s primary daily-driver API SKU at $4.00 input / $20.00 output per MTok (1M context, 128K max output), with $0.20/MTok cache reads (5% of base input), $5/$8 cache writes, $2/$10 batch pricing, and $8/$40 fast mode (API pricing).
TAKEAWAY PDF & MARKDOWN TWIN
Claude Opus 5.5 Cuts Token Price 20% — Takeaway PDF
Anthropic launched Claude Opus 5.5 on September 22, 2026. Seat prices on claude.com did not move. The operator event is the unit price: Opus 5.5 lists at $4 per million input tokens and $20 per million output tokens — 20% below Opus 5 at $5 / $25.
Opus 5.5 is a new SKU, not a rename of Opus 5. Anthropic noted at launch that it matched prior Fable 5.1 on most work (Fable 5.2 subsequently launched at $8/$40 with $0.20 cache reads) and costs about 40% less than Opus 5 on typical token-billed workloads because cache reads fell more than the headline rate.
Opus 5
Opus 5.5
Input / MTok
$5
$4
Output / MTok
$25
$20
5-minute cache write
$6.25
$5
1-hour cache write
$10
$8
Cache read
$0.50
$0.20
Fast mode
$10 / $50
$8 / $40
Batch
$2.50 / $12.50
$2 / $10
Cache reads on Opus 5.5 are 0.05x input ($0.20), not the 0.1x used on Opus 5. That is the line that moves agent and Claude Code bills. US-only inference stays 1.1x on both SKUs.
What did not change
Free $0. Pro $20 monthly / $17 annual. Max $100 (5x Pro). Ultra $200 (10x Pro weekly). Team Standard $20 annual / $25 monthly per seat. Team Premium $100 annual / $125 monthly. Enterprise $20/seat plus usage at API rates. Current sibling lineup (all 1M context / 128K output): Fable 5.2 lists at $8 / $40 ($0.20 cache reads, superseding legacy Fable 5.1 at $10 / $50 and $0.25 cache reads), Sonnet 5.5 lists at $2 / $10 ($0.10 cache reads), and Haiku 5.5 lists at $0.10 / $0.50 for prompts up to 100k tokens ($0.50 / $2.50 above 100k); legacy Haiku 4.5 remains $1 / $5 (200K context).
Opus 5.5 is available on Pro, Max, Ultra, Team, and Enterprise. Anthropic also said five-hour usage limits on those plans increased with the launch. Confirm the live window in Settings > Usage. Do not treat a help-center message count as a hard cap.
Operator note
If you are pinned to claude-opus-5, you are still on the $5 / $25 meter. Switch to claude-opus-5-5 if the workload can move. Fast mode is first-party Claude API only at $8 / $40. Cursor on-demand that routes to Anthropic list will pick up the new SKU when Cursor exposes it — Cursor seat prices on cursor.com ($20 Individual, $40 Teams) did not move with this launch.
The live Anthropic platform table lists Claude Sonnet 5 at $2 input / $10 output per million tokens (as named on the September 22 table — the current model at these same rates is Sonnet 5.5). Five-minute cache writes are $2.50. Cache reads are $0.20. Batch is $1 / $5.
That is not Sonnet 4.6. Sonnet 4.6 (and Sonnet 4.5) — both legacy/retired models — stayed at $3 / $15. Do not average the two rows.
On September 17 this desk published that Sonnet 5 intro $2 / $10 had ended August 31 and that standard list was $3 / $15. The official table on September 22 does not match that desk. Official page wins. Old → new on the live pages: Sonnet 5 $3 / $15 → $2 / $10.
Seats did not move: Pro $20 monthly / $17 annual, Max 5x $100, Max 20x $200, Team Standard $20 annual / $25 monthly, Team Premium $100 annual / $125 monthly, Enterprise $20/seat plus API usage.
Fable 5.1 remains $10 / $50 with cache reads at $0.25 (already logged September 19). Cursor individual seats remain Hobby free / Pro $20 / Pro+ $60 / Ultra $200.
Updated in place: the four canonical pricing desks plus the Claude Code vs Cursor page.
Correction — October 2, 2026: Anthropic reversed course. Sonnet 5 did not move to $3 / $15 on September 1 — it remains $2 / $10 on the first-party API table (batch $1 / $5), and Sonnet 5.5 is the current Sonnet at the same $2 / $10. The September 17 operator note below is preserved as the record of what the card showed then.
Sonnet 5 is no longer $2 / $10 on the first-party API table. Anthropic published introductory pricing through August 31, 2026, then the standard rate: $3 per million input tokens and $15 per million output tokens starting September 1, 2026.
That is a 50% unit-price move on input and output versus the intro card. Sonnet 4.6 was already $3 / $15. The two mid-tier SKUs now share the same list. Cache writes follow the new input rate ($3.75 / MTok for 5-minute writes; $0.30 / MTok cache reads). Batch stays 50% off that list.
What did not move on this check:
Claude Pro $20/mo or $17/mo annual ($200 prepaid)
Max 5x $100/mo, Max 20x $200/mo
Team Standard $25 monthly / $20 annual; Premium $125 / $100
Lineup currency (Oct 2026): Current API list (Oct 2026, verified): Sonnet 5.5 $2/$10, Opus 5.5 $4/$20, Haiku 4.5 $1/$5, Fable 5.1 $10/$50. Legacy (still listed on Anthropic’s card): Opus 4.8 $5/$25, Sonnet 4.6 $3/$15.
Cursor Hobby free, Pro $20, Pro+ $60, Ultra $200, Teams $40/user
If a workflow was budgeted during the September scare, the meter is back to $2 / $10 — the same as Sonnet 5.5. Haiku 4.5 is still the cheap public list. Live desks: API rates, seats.
Anthropic unified standalone Claude Cowork into core Claude on September 16, 2026 (“Claude Cowork is now just Claude”). Multi-step task decomposition, sub-agent delegation, and desktop workspace workflows now run directly inside integrated Claude on Pro ($20/mo), Max ($100/mo), Ultra ($200/mo), Team, and Enterprise plans.
Direct Answer (November 13, 2026):Standalone Claude was merged into core Claude on September 16, 2026. Inside integrated Claude, the desktop workspace mode (formerly Claude) runs directly in the Claude Desktop app on approved local folders and Windows/macOS apps, alongside cloud-hosted Scheduled Tasks (/schedule). Use quick chat for conversation; use Claude Desktop workspace when the job requires local folders, documents, or multi-step workflows. Available on Pro ($20/mo), Max ($100/mo), Ultra ($200/mo), Team, and Enterprise — not Free.
Claude lets Claude work on the machine, not just talk about the work. Last verified November 13, 2026.
Quick Chat vs. Desktop Workspace vs. Chrome Extension
Chat — conversation, no local workspace.
Claude — local files, Skills, desktop workflows in the Claude app.
Chrome extension — the page in the browser. Separate install. See Claude in Chrome.
Claude missing HCS services — the local helper stack did not start. Quit the Claude app, confirm disk space, relaunch. If it persists, reinstall the desktop app. Do not treat this as a chat usage-limit.
Failed to start Claude’s workspace / not enough disk space — free space on the volume that holds the workspace, then retry. Claude and Code spin their own copy of the server.
Couldn’t start this server for Claude and Code sessions — same class. The session runner failed to boot. Restart the app; do not keep retrying from Chat.
Claude not working — confirm you are in the desktop app on a paid seat, not claude.ai Chat and not the Chrome side panel.
How Claude works
How Claude mode works.
In the desktop app Claude can read folders you grant, write documents, run Skills, and call MCP tools. Actions ask for approval. It is not a silent screen-scraper.
What it can automate
File management, document creation (DOCX, XLSX, PPTX, PDF), data cleanup, and cross-app handoffs you approve. It is not a replacement for Claude Code on a repo.
Claude vs Claude Code
Code is the terminal path for developers. Claude is the visual path for operators. Both sit on Pro ($20/mo) / Max ($100/mo) / Ultra ($200/mo) / Team / Enterprise. Matrix: Claude vs Code vs Agent SDK.
Getting started
Install the Claude desktop app. Open a Claude Desktop session. Grant the folder. Describe the job. Approve the first file write before you let it run a batch.
From Research Preview to Integrated Claude: Key Milestones (2026 Timeline)
Readers arriving from our earlier changelog and GA coverage should note three milestones that shaped Claude’s desktop automation surface:
January–March 2026 (Preview Hardening & Computer Use): Stability hardening reduced mid-task stream drops and sub-agent MCP failures; Pro and Max subscribers gained desktop computer use and Dispatch queue support, while power users hit the local VM sessiondata.img limit (GitHub #30751).
April 29, 2026 (General Availability on macOS & Windows): Anthropic removed the research-preview label and added Enterprise OpenTelemetry observability, Usage Analytics API hooks, role-based access controls (RBAC), persistent mobile-to-desktop threads, the unified Customize panel (skills, plugins, connectors), and UI-managed scheduled tasks.
September 16, 2026 (Unification Into Core Claude): Anthropic retired the standalone “Cowork” tab and merged its agentic planner, Skills, MCP tools, and desktop workspace directly into core Claude across Pro ($20/mo), Max ($100/mo), Ultra ($200/mo), Team, and Enterprise.
Frequently Asked Questions
What is Claude’s desktop workspace (formerly Claude)?
Desktop workspace in the Claude app for files and local workflows.
Direct Answer (November 13, 2026):Claude Pro costs $20/month ($17/month billed annually) and covers normal daily work ($0 API credits; Fable via usage credits). Claude Max costs $100/month for 5× Pro usage, $100/month in API credits, and Fable 5.2 up to 50% of your weekly pool. Claude Ultra costs $200/month (formerly Max 20×) for 10× Pro weekly usage, no 5-hour session cap for verified accounts, $200/month in API credits, and Fable 5.2 up to 50% weekly.
Claude Pro costs $20/month. Claude Max costs $100/month; Ultra costs $200/month. The question everyone asks: is 5x or 10x the price actually worth it? The answer depends entirely on how you use Claude. This comparison breaks down the real differences — not the marketing bullet points — so you can make an informed decision.
Pro at $20/month ($17/month annual) includes Claude Code, integrated desktop/agent workflows (Cowork was folded into core Claude), unlimited Projects, Research mode, access to additional models, and Claude for Microsoft 365 and Outlook. The usage allowance is described as “more usage” compared to Free — in practice, this means you can have sustained conversations throughout a workday without hitting limits under normal use. Most professionals who use Claude as a daily tool — a few hours of active conversation per day — find Pro sufficient.
What Max Adds
What Max adds.
Max and Ultra are separate plans. The $100/month tier gives approximately 5x the usage of Pro. The $200/month Ultra tier gives 10x Pro weekly usage. Beyond the usage multiplier, Max adds three concrete features: higher output limits for all tasks (longer responses, more complex code generation), early access to advanced Claude features before they reach Pro users, and priority access during high-traffic periods (you skip the queue).
Direct answer:Max ($100/mo) and Ultra ($200/mo) are the individual plans that include Anthropic’s flagship Fable model. On Pro, Fable usage runs on pay-as-you-go usage credits; on Max and Ultra, Fable is included up to half the weekly usage pool. If your work benefits from Fable — long-horizon agent loops with heavy cache reuse — this alone can decide the plan. Current flagship Fable 5.2 runs $8 input / $40 output per MTok with $0.20 cache reads (down from Fable 5.1’s $10/$50 and $0.25).
Who Actually Needs Max
Who actually needs Max.
Heavy Claude Code users: If you spend 4+ hours per day actively using Claude Code for development — not just coding, but running extended agent sessions, multi-file refactoring, and complex debugging — you’ll likely hit Pro limits. Max ($100/mo) removes this friction. Content production teams: If you’re producing 10+ pieces of content per day through Claude, the usage adds up fast. Researchers and analysts: Extended research sessions with multiple deep-dive conversations consume significant tokens. Anyone who hits Pro limits regularly: If you see the “usage limit reached” message more than once or twice per week, Max pays for itself in recovered productivity.
Who Should Stay on Pro
Most individual professionals: If you use Claude for 1-3 hours per day with normal conversation patterns, Pro is plenty. Occasional users: If Claude is one of many tools in your workflow rather than the central one, Pro is more than enough. Budget-conscious users: At $20/month, Pro delivers extraordinary value. The jump to $100/month should be justified by measurable productivity gains. Users who haven’t hit Pro limits: If you’ve never seen the usage limit message, you don’t need Max.
When you hit the wall: usage credits
One newer mechanic worth knowing before you upgrade: when a paid plan hits its usage limit, usage credits keep Claude working at standard API rates instead of making you wait for the 5-hour reset window. Credits are pay-as-you-go on top of the subscription — the middle path for spiky weeks where one heavy project doesn’t justify a whole tier upgrade. If you hit Pro limits twice a month rather than twice a week, credits are usually cheaper than Max.
The Math on Max
Max ($100/mo) costs $80/month more than Pro. If that additional usage saves you 2+ hours per week of productivity (waiting for limits to reset, or time spent on tasks you could have delegated to Claude), and your time is worth $40+/hour, Max pays for itself. Ultra at $200/month ($180 more than Pro) needs to save roughly 4.5 hours/week to break even at $40/hour. The early access and priority features are hard to quantify financially — they matter most for users who need Claude reliably during peak demand.
A Better Strategy Than Max
Before upgrading to Max, consider whether your usage patterns can be optimized on Pro. Use Projects to maintain context instead of repeating background in every conversation. Be concise in prompts — verbose prompts consume more tokens. Use the appropriate model (Haiku for simple tasks, Sonnet for standard work, Opus for complex reasoning). Close conversations you’re done with rather than continuing indefinitely. If you’re hitting limits despite these optimizations, Max is the right move.
How Claude usage limits actually work
Claude usage limits reset on a rolling five-hour window, and paid plans add weekly limits on top of that. There is no fixed number of messages.
On claude.com/pricing, checked November 5, 2026, Free covers everyday questions. Pro includes at least 5x more usage per five-hour session than Free. Max includes 5x Pro usage; Ultra provides 10x Pro weekly usage with no 5-hour limit for verified accounts. Max costs $100 per month with 5x Pro usage; Ultra costs $200 per month with 10x Pro weekly usage. Chat on the web, the desktop app, mobile, and Claude Code all draw from the same pool.
What spends the allowance is the work itself. A longer conversation, a heavier model, file creation, or research uses more of the window than a short question. Anthropic also says it may apply weekly or monthly caps, or limits on a specific model or feature, so one account does not crowd out everyone else. On the plan table, Fable is unavailable on Free, listed as usage credits on Pro, and listed as 50% of weekly limits on Max and Ultra.
When a limit hits, the same page gives three moves: wait for the reset, move to a higher plan, or, on a paid plan, turn on usage credits and keep working at standard API rates. Settings, then Usage, shows the current status. Team seats use a different meter. A standard seat includes more usage than Pro, at $20 per seat per month billed annually or $25 billed monthly. A premium seat includes 5x the usage of a standard seat, at $100 per seat per month billed annually or $125 billed monthly.
Stay on Pro ($20/mo) unless you hit daily usage caps or need Opus/Fable throughput; grab the one-page PDF cheat sheet or check live Anthropic pricing directly.
A Claude API bill is input tokens plus output tokens, and output costs more than input on every current model. Prices below are US dollars per million tokens. Pricing checked November 13, 2026.
The official table also lists older and retired models. Those rows are omitted here. Sonnet 5 is the same $2 input and $10 output as Sonnet 5.5, and Anthropic says the increase to $3 and $15 that had been scheduled for September 1, 2026 will not happen.
The model and the length of the reply do most of the damage. Output is five times the input price on each model above, so an agent that writes a long diff costs more than one that only reads. Three other switches change the same rates. The Batch API takes 50% off input and output when the job can wait: Fable 5.2 batch is $4 and $20 (legacy Fable 5.1 is $5 and $25), Opus 5.5 is $2 and $10, Sonnet 5.5 is $1 and $5, and Haiku 5.5 is $0.05 and $0.25. US-only inference on Claude 4.6 and later multiplies input, output, cache writes, and cache reads by 1.1. Fast mode, still a research preview, prices Opus 5.5 at $8 input and $40 output, and Opus 5 or Opus 4.8 at $10 and $50. Fast mode does not run on the Batch API.
Claude 4.6 and later include the full 1 million token context window at that same per-token rate. A 900,000-token request is billed at the same rate per token as a 9,000-token request. Claude 4.7 and later, and Mythos Preview, use a newer tokenizer that produces about 30% more tokens for the same text, so an unchanged document can cost more after a model move. Sonnet 4.6 and earlier use the previous tokenizer.
Tools add their own lines. The pricing page lists web search at $10 per 1,000 searches, separate from the tokens used to process the request. Code execution includes 50 free hours a day per organization, then $0.05 per hour per container.
Prompt caching saves money when the same long prefix is sent again inside a short window. You pay extra to store that prefix, then a fraction of the normal input price each time a later request reads it.
Anthropic’s docs, checked November 13, 2026, price a 5-minute cache write at 1.25 times the base input price and a 1-hour cache write at 2 times. A cache hit on Haiku 5.5 costs 10% of base input ($0.01/MTok ≤100K, $0.05/MTok >100K), while higher-tier reads drop further: Fable 5.2 cache hits are $0.20/MTok (2.5% of its $8 base input; legacy Fable 5.1 is $0.25), Opus 5.5 hits are $0.20/MTok (5% of $4), and Sonnet 5.5 hits are $0.10/MTok (5% of $2). On the standard table, a 5-minute write is $10.00 for Fable 5.2, $5.00 for Opus 5.5, $2.50 for Sonnet 5.5, and $0.125 for Haiku 5.5 (≤100K); a 1-hour write is $16.00, $8.00, $4.00, and $0.20.
The buyers who come out ahead repeat a prefix: a long system prompt, a tool list, a standing document, or an agent loop that reloads the same context every turn. A one-off short chat has nothing to reuse. Cache writes are charged when the prefix is first stored. Cache reads are charged when a later request retrieves it, and a read also refreshes the entry. Those multipliers stack with the Batch API discount and with the 1.1x US-inference multiplier.
At the everyday paid chat tier, Claude Pro and ChatGPT Plus are both $20 a month. Above that, compare the usage multiple on the plan, and price the API separately from the subscription.
Checked November 13, 2026. Claude lists Free at $0, Pro at $20 a month or $17 a month when $200 is billed up front for the year, and Max from $100 a month with 5x Pro usage, and Ultra at $200 a month with 10x Pro weekly usage. OpenAI’s Help Center lists ChatGPT Plus at $20 a month, billed monthly, and two Pro tiers: $100 for 5x the usage of Plus, and $200 for 20x the usage of Plus. The Plus article says API usage is billed separately. ChatGPT Go appears as a lower plan; a current Go dollar price was not on the pages read for this check.
The meters differ even when the monthly price lines up. Claude draws web, desktop, mobile, and Claude Code from one pool, on a rolling five-hour window, with weekly limits added on paid plans. OpenAI’s Pro article says models can have separate allowances, and reaching one model’s allowance does not by itself end the plan. Pick the product whose limit matches the work.
API rates below are the standard paid tier, in US dollars per million tokens, from our verified November 2026 check. Where a vendor publishes a short-context rate and a long-context rate, both are included.
Claude Haiku 5.5 is $0.10 input and $0.50 output (up to 100k prompts). Sonnet 5.5 is $2 and $10. Opus 5.5 is $4 and $20. Fable 5.2 is $8 and $40 (legacy Fable 5.1 is $10 and $50). Claude 4.6 and later keep that per-token rate through a 1 million token context window.
On OpenAI’s standard tab, gpt-6-luna is $0.10 input and $0.50 output for short context, and $0.20 and $0.75 for long context. gpt-6.1-sol is $2 and $10 short, and $4 and $15 long. gpt-6-astra is $10 and $50 short, and $20 and $75 long. The same page also publishes Batch, Flex, Fast, and Ultrafast rates.
On Google’s Gemini API, standard paid, Gemini 3.8 Flash is $0.75 input and $3.75 output through December 31, 2026, then $1.50 and $7.50. Gemini 3.1 Pro Preview is $2 input and $12 output on prompts up to 200,000 tokens, and $4 and $18 above that. Gemini’s output price includes thinking tokens.
The useful comparison is the shape of the bill. On all three, output tokens cost more than input tokens. OpenAI’s flagship models and Gemini 3.1 Pro Preview raise the per-token rate once a prompt crosses their long-context line. Claude’s current models keep one rate inside the 1 million token window. A repeated prefix is discounted on each platform under that platform’s own cache rules.
What is the difference between Claude Pro and Claude Max?
The primary difference between Claude Pro and Claude Max is capacity, monthly usage headroom, and bundled developer credits on identical underlying models. Anthropic prices Claude Pro at $20 per month (or $17 per month when billed annually at $200 upfront). Claude Max is priced at $100 per month, delivering an official 5x usage multiplier over Pro across both the 5-hour rolling session window and the weekly account cap.
Beyond session headroom, Claude Max includes $100 per month in Claude Platform API credits, priority access during peak network traffic times, early access to new Claude features, and included access to Anthropic’s flagship Fable model up to half your weekly allowance. For extreme workloads, Anthropic also offers an Ultra tier at $200 per month, providing 10x weekly usage, $200 per month in API credits, and no 5-hour limit for verified accounts.
Workload fit determines the right tier: Claude Pro fits solo professionals, writers, and analysts working 1 to 3 hours per day with moderate prompt lengths. Claude Max fits full-time developers, engineering leads running continuous Claude Code sessions, and power users who regularly hit Pro’s 5-hour rolling limits mid-task.
How much does Claude Pro cost?
Claude Pro costs $20 per month when billed monthly, or $17 per month when billed annually ($200 billed upfront). Anthropic’s published terms confirm that Pro provides substantially more usage than the Free tier across both the 5-hour session window and weekly cap, access to advanced models (Sonnet and Opus), Projects for context management, Claude Code, Claude Science, Claude Dashboards (beta), and priority bandwidth during high-traffic periods.
Is Claude Max worth it for daily coding?
For daily coding, Claude Max is worth the investment if you write software full-time and frequently run into Pro’s session-based rate limits. In daily developer workflows involving continuous terminal interactions, multi-file diffs, and iterative debugging in Claude Code, each prompt re-submits previous code turns and file context. On Claude Pro, intensive coding sessions often hit the 5-hour rolling cap within 60 to 90 minutes of active development.
The financial math on Max for coding is straightforward: Max costs $100 per month, an $80 monthly premium over Pro. That plan includes an official 5x usage multiplier across 5-hour and weekly windows, plus $100 per month in API credits to build on the Claude Platform. If avoiding interrupted coding sessions saves you just 2 hours per week of downtime, and your professional developer time is valued at $40 or more per hour, Claude Max pays for itself within the first week.
What does pay-as-you-go API usage beat a Max subscription?
Pay-as-you-go API usage beats a Claude Max subscription when your workloads are automated, bursty, or integrated directly into developer environments rather than open-ended conversational web chats. Instead of paying a flat $100 or $200 monthly subscription fee for personal seat headroom, direct API access bills only for the exact tokens your tasks consume at standard published rates (such as Claude Sonnet 5.5 at $2.00 per million input tokens and $10.00 per million output tokens, or Haiku 5.5 at $0.10 input and $0.50 output).
For programmatic scripting, IDE extensions, or scheduled batch jobs, typical developer consumption often totals $15 to $40 per month in API charges—well below the $100 per month cost of a Max seat. Furthermore, API usage unlocks prompt caching (reducing input token costs up to 90% on cached prefix reads) and Message Batches (providing an automatic 50% discount for asynchronous turnaround within 24 hours). For a detailed structural analysis of when to buy per-user seats versus consuming raw tokens, see our analysis of Seats vs API — the question models get wrong.
Frequently Asked Questions
Is Claude Max worth $100 a month?
If you regularly hit Pro usage limits — especially heavy Claude Code users, content producers, or researchers — Max pays for itself in recovered productivity. If you’ve never hit Pro limits, stay on Pro.
What is the difference between Max and Ultra?
Max ($100/month) gives 5x Pro usage. Ultra ($200/month) gives 10x Pro weekly usage with no 5-hour limit for verified accounts. Both include higher output limits, early feature access, and priority. Most users who need Max find 5x sufficient.
Can I switch between Pro and Max?
Yes. You can upgrade from Pro to Max or downgrade from Max to Pro at any time. Changes take effect on the next billing cycle.
Does Max include Claude Code?
Yes. Both Pro and Max include Claude Code. Max gives you more usage capacity for Claude Code sessions.
Does Claude Max include Fable?
Yes — both Max ($100/mo) and Ultra ($200/mo) include Fable models up to 50% of the weekly usage pool. On Pro, Fable usage runs on pay-as-you-go usage credits.
What are Claude usage credits?
Pay-as-you-go tokens billed at standard API rates that keep you working when a paid plan hits its limit, instead of waiting for the reset window. Ideal for occasional spike weeks.
Can I pay for Max or Ultra annually?
No. Max and Ultra are monthly billing only. Pro is the plan with an annual option — $17/month billed yearly ($200 up front).
The Next Step · Research Verdict
Which Claude path makes sense for you?
Most individual professionals and developers working 1–3 hours daily should stay on Claude Pro ($20/mo or $17/mo annual) — it includes Claude Code, Projects, and generous throughput. The $100/mo Claude Max upgrade only pays for itself if you hit Pro rate limits daily inside Claude Code or require Claude Fable reasoning models.
Updated November 13, 2026 — facts, pricing tiers, models, and field-sheet PDF verified against live Anthropic documentation.
Direct Answer · Verified November 13, 2026
Claude scales editorial, SEO, email, and social workflows by pairing detailed content briefs with brand-voice Projects — drafting structured H2/H3 outlines, on-page optimization passes, nurture sequences, and social repurposing while leaving fact-checking and original insight to human editors (claude.com/product/overview). Throughput scales with editorial depth and 5-hour/weekly plan caps rather than a fixed daily article quota: solo writers start on Pro ($20/mo, $17 annual), high-output operators step up to Max ($100) or Ultra ($200), and multi-writer teams run Team Standard ($20 annual / $25 monthly). Compare tiers on Claude AI pricing and pair team playbooks with solo creator workflows on Claude for content creators.
SEO & CONTENT FIELD SHEET
Claude for Content Creation & SEO — Tygart Media Field Sheet (PDF)
Consolidated content, SEO & marketing team playbook: 400–600-word brief pipelines, competitive synthesis, A/B copy variants, and Pro/Team/Ultra usage specs. Verified 2026-11-13.
Claude has become a core tool for content teams — not as a replacement for human writers, but as a force multiplier that changes what’s possible with limited resources. This guide covers the practical workflows that professional content creators, SEO specialists, and marketing teams use with Claude in 2026, including where it excels, where it falls short, and how to integrate it into a production content operation.
Blog Posts and Long-Form Content
Blog posts and long-form with Claude.
Claude excels at drafting long-form content when given proper direction. The key is providing a detailed brief — not just a topic, but the target keyword, the audience, the desired structure, the tone, competing content to differentiate from, and any specific data or examples to include. A well-briefed Claude request produces a first draft that’s 70-80% of the way to publishable, versus a vague request that produces generic filler.
Best practices for blog production: write a content brief first (or use Claude to help write one), include your brand voice guidelines in a Project, specify the exact structure you want (H2s and H3s), request specific word counts, and always edit the output for accuracy, originality, and brand alignment. Never publish AI-generated content without human review — this is especially important for factual claims, statistics, and technical accuracy.
SEO Content Optimization
SEO content optimization workflows.
Claude can analyze existing content for SEO improvements — identifying missing keywords, suggesting heading structure changes, improving meta descriptions, and recommending internal linking opportunities. Feed Claude your target keyword, your current content, and competitor content, and ask for specific optimization recommendations. Claude can also generate FAQ sections with structured data markup, which directly targets featured snippets and People Also Ask placements.
For new content, Claude can research keyword clusters, identify search intent, and draft content structured for both traditional SEO and emerging AI search optimization (AEO/GEO). The combination of web search capability and content generation means Claude can research a topic and draft optimized content in a single session.
Email Marketing
Claude handles email marketing content effectively — subject line variations, body copy, CTAs, and nurture sequences. The workflow that works best: share your product/service details and audience information in a Project, then request specific email types (welcome sequence, promotional, re-engagement, newsletter). Claude can generate multiple variations for A/B testing and adapt tone for different segments.
Social Media Content
Claude can repurpose long-form content into social media posts tailored for different platforms — LinkedIn articles and thought leadership posts, Twitter/X threads, Instagram captions, and Facebook updates. Provide the source content and specify the platform, tone, and any hashtag or formatting requirements. Claude adapts naturally between professional (LinkedIn), conversational (Twitter), and visual-caption (Instagram) styles.
Content Strategy and Planning
Content strategy and planning.
Beyond individual pieces, Claude can help with content strategy — editorial calendar planning, content gap analysis, persona development, and competitive content auditing. Upload your existing content inventory, share your business goals and target audience, and ask Claude to identify gaps, suggest topics, and prioritize based on potential impact. This is especially powerful with web search enabled, allowing Claude to analyze competitor content in real-time.
Marketing Team Pipelines, Synthesis & Workspace Setup
Marketing teams get the strongest leverage from Claude not as an autonomous content button, but as a structured thinking partner and first-draft engine that eliminates the repetitive 80% of structural work while leaving strategy, positioning, and final editorial judgment with humans.
Start every assignment inside a Brand Voice Project with a structured 400–600-word creative brief before asking for prose so writers spend their review cycle sharpening accuracy and angle rather than building scaffolds. For campaign launches, supply your objective, target buyer, budget tier, and competitive context to generate structured messaging frameworks and launch briefs your whole marketing team can critique and align around.
Competitive Research Synthesis & A/B Copy Variants
Feed Claude a competitor’s landing page alongside reference articles and your current positioning to surface messaging gaps and untested angles before creative kickoffs. For paid acquisition and lifecycle testing, provide your control copy and request five or more testable variants built around distinct hooks, CTAs, and tone registers — letting strategists pick the top two or three candidates for live split tests.
Organizing Shared Projects Across a Marketing Team
Deploy shared channel Projects — separate Projects for blog, email, social, and paid creative, each preloaded with voice guides, buyer personas, and past top performers — on Team Standard ($20/seat/mo annual or $25 monthly; 2–150 seats) with SSO, central billing, and zero model training by default. Heavy campaign producers can mix in Team Premium ($100 annual / $125 monthly, 5× Standard capacity) inside the same workspace, while organizations with 20+ seats needing SCIM, audit logs, and compliance API controls deploy Enterprise ($20/seat/mo annual + API usage).
Quality Control and Accuracy
AI-generated content requires human quality control. Every piece should be checked for factual accuracy (especially statistics, dates, and specific claims), brand voice consistency, originality (run through plagiarism detection), legal compliance (disclaimers, disclosures), and genuine value to the reader. The biggest risk with AI content is not that it’s bad — it’s that it’s competent but generic. Human editors should push for the specific insights, examples, and perspectives that make content genuinely useful rather than just technically correct.
Yes. Claude can draft keyword-optimized content, generate meta descriptions, create FAQ sections with schema markup, and analyze content for SEO improvements. Human review for accuracy and originality is essential.
Should I use Claude to write my entire blog?
Use Claude as a drafting and optimization tool, not a hands-off content factory. The best results come from human-directed Claude drafts that are then edited for accuracy, brand voice, and genuine insight.
Can Google detect AI-written content?
Google has stated it focuses on content quality regardless of how it’s produced. The key is creating content that’s helpful, accurate, and provides genuine value — whether written by humans, AI, or both.
How much content can Claude produce per day?
Throughput depends on research depth, primary-source verification, and editing rigor rather than a fixed daily article quota. On Pro ($20/mo, or $17/mo annual), subscribers receive at least 5× Free usage per 5-hour session plus a weekly cap across current 1M-context models (Sonnet 5.5, Opus 5.5); high-volume operators step up to Max ($100/mo), Ultra ($200/mo, no 5-hr limit for verified accounts), or Team Standard ($20/seat annual / $25 monthly) — see full limits on our Claude AI pricing guide and workspace specs on claude.com/product/overview.
The Next Step · Verified Research Verdict (November 13, 2026)
Next Step: Standardize Brief Templates & Align Team Plans
Claude for Content Creation & SEO — Tygart Media Field Sheet (PDF)
Download the consolidated Content, SEO & Marketing Teams Field Sheet (Verified 2026-11-13) for brief-first pipelines and Team governance. Verified 2026-11-13.
Model Context Protocol (MCP) is an open standard created by Anthropic that lets Claude connect to external tools, data sources, and services. Instead of copying data into Claude manually, MCP gives Claude structured access to the tools you already use — databases, APIs, project management platforms, file systems, and more. MCP has become one of the most important developments in the AI ecosystem in 2026, and understanding it is increasingly essential for developers and technical teams.
What MCP Actually Does
What MCP actually does — hosts, clients, servers.
At its core, MCP is a protocol — a standardized way for AI models to communicate with external services. Think of it like how HTTP standardized web communication or how SQL standardized database queries. MCP standardizes how AI assistants request and receive data from external tools. Before MCP, connecting Claude to a database required custom integration code. With MCP, you configure an MCP server that speaks the protocol, and Claude can query the database through that server using a standardized interface.
The Architecture: Hosts, Clients, and Servers
Architecture: hosts, clients, and servers.
MCP has three components. The host is the application where Claude runs (the desktop app, Claude Code, or a custom application). The client is the MCP client built into Claude that manages connections to MCP servers. The server is the service that provides tools, data, or capabilities to Claude. MCP servers expose three types of primitives: tools (actions Claude can take, like querying a database or creating a Jira ticket), resources (data Claude can read, like file contents or documentation), and prompts (pre-built interaction patterns).
Practical Examples
A Notion MCP server lets Claude read and write Notion pages and databases directly. A PostgreSQL MCP server lets Claude query your database. A Slack MCP server lets Claude read channels and send messages. A GitHub MCP server lets Claude interact with repositories, issues, and pull requests. A Sentry MCP server lets Claude access error tracking and debugging data. These aren’t hypothetical — they’re production tools that teams use daily.
Local vs Remote MCP Servers
MCP servers can run locally on your machine or remotely as hosted services. Local MCP servers run alongside the Claude desktop app and have access to your local environment — file system, local databases, development tools. They use the stdio transport (standard input/output) and require no network configuration. Remote MCP servers run as web services and are accessed over the network using Streamable HTTP or Server-Sent Events (SSE) transports. Remote servers can be shared across teams and don’t require local installation.
Token Cost Considerations
An important practical consideration: MCP tools add tokens to every conversation turn. Each configured MCP server’s tool descriptions are included in Claude’s context, consuming input tokens. If you have 10 MCP servers with 5 tools each, that’s 50 tool descriptions included in every request — potentially thousands of tokens per turn. Best practices include only connecting the MCP servers you actively need, using scoped configurations to limit which tools are available in which contexts, and monitoring your token usage to identify MCP-related costs.
Why Developers Care
Why developers care — less custom glue.
MCP matters because it transforms Claude from a standalone chatbot into a connected agent. Without MCP, Claude can only work with information you paste into the conversation. With MCP, Claude can pull real-time data, take actions in external systems, and operate as part of your existing toolchain. For development teams, MCP means Claude Code can interact with your entire development stack — version control, CI/CD, error tracking, documentation, project management — through a single standardized interface.
Getting Started with MCP
The fastest path is to install a pre-built MCP server for a tool you already use. The Claude desktop app’s settings include MCP server configuration. Add a server definition (the server command and its arguments), restart Claude, and the tools become available in your conversations. For custom integrations, Anthropic provides SDKs for building MCP servers in Python and TypeScript. The MCP specification is open — anyone can build a server for any tool.
MCP (Model Context Protocol) is an open standard that lets Claude connect to external tools and data sources — databases, APIs, file systems, and more — through a standardized interface.
Is MCP free to use?
MCP itself is free and open. MCP servers may be free (open source) or paid (commercial). The token costs from MCP tool descriptions are included in your regular Claude usage or API billing.
Do I need to be a developer to use MCP?
Basic MCP server setup requires some technical comfort — editing configuration files and running commands. Pre-built connectors in the Claude interface are simpler. Building custom MCP servers requires programming knowledge.
Can MCP be used with other AI models?
MCP is an open protocol. While Anthropic created it for Claude, other AI platforms and tools have begun adopting MCP as a standard for tool integration.
The Signal — Tygart Media's weekly brief. What we actually ran, what it measured, and what it means for a restoration shop.