Last Updated: 21 September 2026

Executive Summary

Claude is Anthropic’s AI assistant, and its reputation rests on writing quality, coding and long-document work. Claude Code is included in every paid plan, the API offers 1-million-token context at standard pricing, and Claude products carry no advertising. The weaknesses are practical rather than hidden. Usage is governed by two stacked limits — a five-hour window and a weekly ceiling — that heavy users hit routinely. There is no native image or video generation. And in June 2026, Anthropic’s newest models were unavailable for 19 days after a US government export-control directive, a reminder that frontier-model access can change without notice.

Estimated reading time: 11 minutes

Key Takeaways

• Pricing runs Free, Pro at $20 monthly or $17 annually, Max at $100 or $200, Team seats from $20, and Enterprise at $20 per seat plus usage billed at API rates.
• Every paid plan includes Claude Code, Anthropic’s coding agent. Free does not.
• Usage is metered by two stacked caps — a rolling five-hour session and a weekly ceiling — shared between chat and Claude Code. This is the most common complaint about Claude.
• On 14 September 2026 a promotional Claude Code allowance ended. The new permanent standard is 25% above the old baseline, which Anthropic itself describes as a 17% reduction compared with the promotional level users had grown used to.
• Claude Fable 5 and Mythos 5 were withdrawn for all customers from 12 June to 1 July 2026 to comply with a US export-control directive, then restored.
• Claude does not generate images or video, which rules it out as a single-subscription replacement for broader assistants.

What Is Claude?

Claude is the AI assistant built by Anthropic. It is available on the web, desktop and mobile, and increasingly across other surfaces: Claude Code for software development, Cowork for knowledge-work tasks on your desktop, Claude in Chrome as a browsing agent, Claude in Excel and Claude in PowerPoint inside Microsoft Office, Claude Design for visual work on a canvas, and Claude Tag, which lets anyone in a Slack workspace delegate tasks to Claude.

The current model lineup has four public tiers: Haiku 4.5 for speed and cost, Sonnet 5 for most production work, Opus 5 for complex reasoning, and Fable 5.1, Anthropic’s most capable publicly available model. Above that sits the Mythos tier. Claude Mythos 5.1 and Fable 5.1 share the same underlying model; Fable carries additional safety measures covering biology, cybersecurity and AI research and development. Claude Mythos Preview is not publicly available and is used by a small number of organisations through Anthropic’s Project Glasswing cybersecurity partnership.

Paid plans differ mainly in how much you can use rather than which models you can access — a meaningful difference from competitors that reserve their best models for higher tiers.

Why this matters for your business: Claude’s pitch is depth rather than breadth. It is aimed at writing, reasoning, coding and document work, not at being a do-everything assistant with images and video attached. Whether that suits you depends on whether your work sits inside that range — and on whether your usage stays inside its limits.

Who Is It Best For?

Claude is a strong fit for:

• Software developers. Claude Code is included from Pro upward, and coding is where Claude is most consistently recommended by independent reviewers.
• Writers, analysts and researchers. Long-form writing, editing and analysis of long documents are core strengths, supported by large context windows.
• Businesses wanting an ad-free assistant with firm data terms. Claude products carry no advertising, and Team and Enterprise exclude training on your data at the contract level.
• Teams already in Slack or Microsoft Office. Claude Tag, Claude in Excel and Claude in PowerPoint bring it into existing tools.
• Developers building on the API. One-million-token context at standard pricing, prompt caching and batch discounts make long-context work economical.

Claude is probably not the right choice for:

• Anyone needing image or video generation, which Claude does not provide.
• Very heavy individual users on Pro, who will hit the stacked usage limits and likely need Max.
• Budget-first individuals — there is no tier between Free and the $17–$20 Pro plan.
• Small organisations wanting Enterprise controls, given a 20-seat self-serve minimum.
• Businesses that cannot tolerate model availability changing at short notice, as happened in June.

Key Features

Claude Code. An agentic coding tool available in the terminal, desktop and web, included in every paid plan. Anthropic’s documentation puts average spend at around $13 per developer per active day when billed at API rates.

Cowork. A desktop agent for non-developers that works across files and applications to complete multi-step knowledge-work tasks.

Office and browser integrations. Claude in Excel, Claude in PowerPoint and Claude in Chrome let it work inside tools people already use rather than in a separate window.

Claude Tag. A Slack integration that lets anyone in a workspace tag Claude into a conversation and hand it a task.

Claude Design. A canvas with design tools where visual work can be iterated through chat.

Long context. Claude 4.6 and later models support one-million-token context at standard pricing, with no long-context surcharge.

Projects and memory. Organised workspaces with shared knowledge, and memory that can carry context across conversations.

The Claude API. Four current models across roughly a tenfold price range, with prompt caching and a 50% batch discount.

How Claude usage is actually metered

EVERY PAID PLAN
      │
      ├─► 5-hour rolling session window
      │        ▲
      │        │  shared across chat AND Claude Code
      │        ▼
      ├─► Weekly ceiling (stacked on top)
      │     Claude Code standard raised 25% on 14 Sep 2026
      │     = 17% less than the promo users had
      │
      └─► Separate monthly credit for Agent SDK,
          headless runs and third-party agents:
          $20 Pro · $100 Max 5x · $200 Max 20x
          $20 Team Standard · $100 Team Premium

[Illustration placeholder: Claude usage-limit structure]

Ease of Use

The core chat experience is simple and uncluttered, and the model lineup is easier to understand than several competitors’: paid plans change how much you can use, not which models you can reach.

The friction is usage limits. Every subscription meters activity against two stacked caps — a rolling five-hour session window and a weekly ceiling — and both are shared between chat and Claude Code. A developer who spends the morning in Claude Code can find chat constrained in the afternoon. This is the most frequently raised complaint about Claude across user reviews, and it is a structural feature of the plans rather than a bug.

September brought a further change. Anthropic ran Claude Code weekly limits 50% above the published standard from 13 May to 13 September 2026. From 14 September, the permanent standard is 25% above the pre-promotion level. That is a genuine increase over the old baseline, but Anthropic’s own framing acknowledges it works out to a 17% reduction compared with what users had during the promotion. Five-hour session limits were not changed.

On paid plans with code execution enabled, Claude automatically summarises earlier messages as a conversation approaches its limit, so long chats can usually continue rather than stopping abruptly.

Future Signal Tip: If you use both chat and Claude Code heavily, watch which hits the limit first. Because the two share the same caps, heavy coding days can quietly consume your chat allowance. If you are regularly interrupted, Max 5x at $100 is usually a better fix than rationing.

Performance

Independent reviewers consistently rate Claude highly for writing, coding and careful reasoning, and Claude Code is widely regarded as one of the leading coding agents. That reputation is corroborated commercially: our earlier review of Anthropic noted that a large share of enterprise AI spending has settled on its models.

Long-context work is a genuine strength. One-million-token context at standard pricing on Claude 4.6 and later removes a common cost trap for contract review, research synthesis and large codebases.

Three honest caveats. First, Claude does not generate images or video, so it cannot replace a broader assistant on its own. Second, benchmark comparisons between frontier models move quickly, and any single result should be treated with caution. Third, developers building on Fable have reported a practical gotcha: refusals can return as a normal successful response rather than an error, which can quietly affect automated pipelines unless outputs are checked.

The Availability and Governance Questions

These deserve more space than a features list allows, and they are presented factually rather than argued.

The June 2026 access suspension. Anthropic released Claude Fable 5 and Claude Mythos 5 on 9 June 2026. On 12 June, Anthropic received a US government export-control directive citing national security authorities, requiring it to suspend access to both models by any foreign national, inside or outside the United States, including its own foreign-national employees. Because it could not verify nationality on every request, Anthropic disabled both models for all customers. Reporting indicated the government’s concern related to a possible method of bypassing safeguards intended to limit Fable 5’s use for certain cybersecurity tasks. A White House adviser publicly said Anthropic had refused to fix the underlying issue; Anthropic disputed that characterisation and described the episode as a misunderstanding. Other models, including Opus 4.8, remained available throughout.

On 30 June, the Department of Commerce lifted the controls, and Anthropic restored access on 1 July. Anthropic’s statement is at anthropic.com/news/fable-mythos-access.

The practical lesson is not political. Law firms advising on the episode noted that businesses which had integrated Fable 5 experienced service interruption even where their own staff raised no access issue, and recommended that API-level integrations be reviewed. Whatever one’s view of the policy, it demonstrated that access to a frontier model can change within hours.

Other matters buyers should know. As covered in our Anthropic review, the company has faced a US Department of Defense supply-chain risk designation that a court blocked in March 2026, a book-piracy copyright settlement of roughly $1.5 billion, and a pending music copyright case. Buyers in government-adjacent sectors should check current status before committing.

On advertising and data. Claude products are ad-free, and Anthropic does not let advertisers pay to have Claude promote products in conversations. Team and Enterprise exclude training on customer data at the contract level.

⚠️ The practical position. Build any production workflow on Claude with a tested fallback model and a documented switch-over process. That advice applies to every frontier provider, but June 2026 made it concrete. For business data, use Team or Enterprise.

Integrations

Claude’s integrations increasingly meet people where they work. Claude in Excel and Claude in PowerPoint operate inside Microsoft Office, Claude in Chrome acts as a browsing agent, and Claude Tag brings it into Slack so anyone in a workspace can delegate a task. Cowork can use these integrations as tools.

On the developer side, the Model Context Protocol — an open standard Anthropic published for connecting AI to external tools and data — has been adopted widely across the industry, including by competitors. That reduces lock-in: connectors built for MCP are not Anthropic-specific.

Claude is also available through Amazon Bedrock, Google Cloud and Microsoft Foundry as well as Anthropic’s own API, which lets organisations buy through existing cloud commitments.

Pricing & Value

Anthropic’s plans are straightforward, and usage headroom is the main thing you pay for.

Max is currently monthly-only on web checkout, with no annual discount. Pro’s annual option is billed as $200 upfront.

API pricing per million input and output tokens: Haiku 4.5 at $1 and $5, Sonnet 5 at $2 and $10, Opus 5 at $5 and $25, and Fable 5.1 at $10 and $50. Cache reads cost a tenth of the input rate and the Batch API halves costs.

⚠️ One pricing conflict to be aware of. Sonnet 5 launched with $2/$10 described as introductory pricing due to rise to $3/$15 on 1 September. Anthropic cancelled that increase on 10 August and made $2/$10 the standard rate, which is what Anthropic’s pricing page and most independent trackers show. At least one third-party guide still lists $3/$15. Rely on Anthropic’s own pricing page.

Two further costs are worth budgeting. Agent SDK usage, headless runs and third-party agents authenticated through a subscription now draw from a separate monthly credit rather than your interactive limits. And, as our Anthropic review noted, newer Claude models use a tokenizer that produces roughly 30% more tokens for the same text, which raises effective API costs at unchanged headline rates.

For context on competitive position: Claude Team and ChatGPT Business currently list identical Standard and Premium seat prices. Claude has no equivalent to ChatGPT’s $8 Go tier. Claude’s Enterprise model — a low seat fee plus usage at API rates — can be cheaper than premium seats for organisations with moderate per-user usage, and more expensive for heavy users.

Is it good value? For developers, Pro is strong value because Claude Code is included. For writers and analysts, Pro covers daily use comfortably. The value question turns almost entirely on whether your usage fits inside the stacked limits.

Strengths

• Consistently strong reviews for writing, coding and long-document work.
• Claude Code included in every paid plan, from $17 a month on annual billing.
• One-million-token context at standard API pricing on current models.
• Claude products are ad-free, and business tiers exclude training at the contract level.
• Paid plans differ by usage rather than locking better models behind higher tiers.
• Open Model Context Protocol and multi-cloud availability reduce lock-in.

Limitations

• Stacked five-hour and weekly usage caps, shared between chat and Claude Code, are the most common complaint.
• Claude Code’s post-promotion limit is 17% lower than the level users had from May to September.
• No native image or video generation.
• Newest models were unavailable to all customers for 19 days in June 2026.
• No budget tier between Free and Pro; Max is monthly-only; Enterprise requires 20 seats self-serve.
• Newer models’ tokenizer raises effective API costs, and ongoing legal matters warrant checking.

Alternatives

ChatGPT — broader in one subscription, including image generation and an $8 entry tier, with identically priced team seats. The stronger choice if you need images or a budget tier.

Google Gemini — very large context windows and deep integration for Google Workspace businesses.

Microsoft 365 Copilot — the default for organisations committed to Microsoft 365, grounded in company data.

Open-weight models — worth considering where data must stay on your own infrastructure or volume is high and tasks are simple.

The Future Signal

Claude’s 2026 direction is clear: move from a chat window into the places work already happens — the terminal, the desktop, the browser, Excel, PowerPoint and Slack — while keeping the product ad-free and the business data terms firm. That is a coherent strategy, and its commercial traction in enterprise suggests buyers are responding to it.

The strategic tension is capacity. Stacked usage limits and the end of a promotional allowance show that demand is running ahead of supply, and every limit change is felt immediately by the developers who rely on Claude most. How Anthropic manages capacity as agentic use grows will shape the user experience more than any single model release.

The second tension is exposure. June’s export-control episode, the defence procurement dispute and ongoing copyright litigation mean Claude carries a more eventful governance record than some competitors. None of it has affected everyday availability of most models, but it is a genuine consideration for regulated and government-adjacent buyers.

Watch two things: whether usage limits loosen as capacity expands, and whether Mythos-class capability becomes more broadly available without further disruption.

Final Verdict

Overall Rating: 8.4 / 10

Claude is an excellent assistant for writing, coding and document-heavy work, and Claude Code’s inclusion in every paid plan is strong value for developers. Claude products are ad-free and business data terms are firm.

It scores below our 8.6 for ChatGPT, and that ordering is deliberate rather than modest: ChatGPT offers broader capability in one subscription, including image generation and a cheaper entry tier. Claude also loses points for stacked usage limits, the September reduction relative to the promotional level, and June’s 19-day withdrawal of its newest models.

Future Signal recommends Claude for developers, writers, analysts and teams whose work centres on text, code and documents — on Team or Enterprise for business data, with a fallback model in place for production workflows. Given the obvious conflict of interest in this review, we strongly recommend testing it against ChatGPT and Gemini on your own tasks.

Before you subscribe — a short checklist

• Confirm your work does not need image or video generation
• Estimate whether your usage fits Pro’s five-hour and weekly caps
• Remember chat and Claude Code share the same limits
• Choose Max 5x if interruptions are frequent, rather than rationing
• Use Team or Enterprise for business and client data
• Keep a tested fallback model for production workflows
• Check API pricing on Anthropic’s own page — some guides are outdated
• Test Claude against alternatives yourself, given this review’s disclosed conflict

Frequently Asked Questions

How much does Claude cost? Free, then Pro at $20 a month or $17 billed annually, Max at $100 or $200 a month, Team Standard at $25 or $20 annually per seat, Team Premium at $125 or $100, and Enterprise at $20 per seat annually plus usage billed at API rates.

Is Claude Code included? Yes, in every paid plan from Pro upward. It is not included on Free. Claude Code draws from the same usage limits as chat.

Why do I keep hitting Claude’s usage limits? Because two limits stack: a rolling five-hour session window and a weekly ceiling, both shared between chat and Claude Code. Heavy coding can consume chat allowance. Max plans raise both.

Did Claude’s limits change in September 2026? Yes. A promotion running Claude Code weekly limits 50% above standard ended on 13 September. From 14 September, the permanent standard is 25% above the old baseline — which Anthropic describes as a 17% reduction compared with the promotional level. Five-hour limits were unchanged.

What happened to Claude Fable 5 in June? Anthropic released it on 9 June 2026. On 12 June it disabled Fable 5 and Mythos 5 for all customers to comply with a US export-control directive. The Department of Commerce lifted the controls on 30 June and access was restored on 1 July. Anthropic’s statement is at anthropic.com/news/fable-mythos-access.

Can Claude generate images? No. Claude does not natively generate images or video. It can create visual work through code and design tools such as Claude Design, but it is not an image generator in the way some competitors are.

Does Claude show ads or train on my data? Claude products are ad-free, and advertisers cannot pay to have Claude promote products. Team and Enterprise exclude training on customer data at the contract level.

Claude or ChatGPT? ChatGPT is broader in one subscription, with image generation and an $8 tier. Claude is often preferred for writing, coding and long documents. Team seat prices are identical. Given this review’s conflict of interest, the honest advice is to test both on your own work.

Conclusion

Claude’s strength is depth in the work it is built for. Writing, coding, reasoning and long-document analysis are consistently rated among the best available, Claude Code is included in every paid plan, and the product is expanding sensibly into the tools where work already happens. It carries no advertising, and its business data terms are firm.

Its weaknesses are equally concrete. Stacked usage limits shared between chat and code are the most common frustration, September’s change left Claude Code users with less than the promotional allowance they had grown used to, there is no image or video generation, and June showed that access to a frontier model can disappear for weeks.

The practical advice is to match Claude to the work. If your needs centre on text, code and documents, it is an excellent choice — on Team or Enterprise for business data, with a fallback in place for anything in production. If you need images, a budget tier or maximum breadth in one subscription, look at ChatGPT first. And because this review was written with Claude’s assistance, test it yourself rather than taking our word for it.