Anthropic Keeps Claude Sonnet 5 at Its Launch Price

30-Second Summary

Anthropic has made the introductory pricing for Claude Sonnet 5 permanent. The model stays at $2 per million input tokens and $10 per million output tokens indefinitely. A 50 per cent increase to $3 per million input and $15 per million output tokens, which was scheduled to take effect on 1 September 2026, will not occur. For businesses using Claude via the API, or evaluating whether to adopt it, today marks the removal of a significant pricing barrier.

At a Glance

• Topic: Model Releases
• Company: Anthropic
• Date: 1 September 2026
• Announcement: Claude Sonnet 5 introductory pricing confirmed permanent on 10 August 2026
• What Changed: A 50 per cent price increase scheduled for today has been cancelled
• Why It Matters: Any business using Claude Sonnet 5 via API avoids a significant cost increase
• Who Should Care: Businesses using the Claude API, teams building AI workflows, organisations evaluating Claude adoption

Key Facts

• Company: Anthropic
• Announcement Date: 10 August 2026
• What Changed: The $2/$10 per million input/output token rate for Claude Sonnet 5 is now the permanent standard price, not a temporary introductory rate
• Who It Affects: All Claude API users and businesses building Claude-powered products or internal tools
• Primary Source: Anthropic platform documentation, confirmed across multiple industry publications

What Happened

Anthropic launched Claude Sonnet 5 with introductory pricing of $2 per million input tokens and $10 per million output tokens. At launch, the company stated this was a temporary introductory rate, with a planned increase to $3 per million input tokens and $15 per million output tokens scheduled to take effect on 1 September 2026.

On 10 August 2026, Anthropic reversed that decision. The company confirmed that the $2/$10 pricing is now the permanent standard rate for Claude Sonnet 5. The September 1 increase will not occur.

The change was confirmed across Anthropic's platform documentation and has been independently verified by multiple industry sources including Enterprise DNA, Explainx.ai, and Finout. The move locks in pricing that is 33 per cent lower on input tokens and 33 per cent lower on output tokens compared to what customers had originally been told to expect from today.

The reversal came against a backdrop of intensifying competition across the AI model market. OpenAI cut pricing on its Luna model by 80 per cent in late July 2026, and Meta released the open-weight Muse Glimmer model under an Apache 2.0 licence in August, allowing commercial use at no cost. Anthropic's decision to lock in competitive pricing reflects a broader market dynamic where model providers are competing on cost as well as capability.

Why It Matters

• Businesses that built cost models assuming the September 1 increase now have a direct budget saving without any action required on their part
• Any organisation that paused Claude-powered workflow projects because of the anticipated price hike can now move forward with stronger economics
• The $2/$10 rate positions Claude Sonnet 5 competitively against comparable mid-tier models from OpenAI and Google at their current market prices
• The pricing lock provides planning certainty for operators who want to commit to Claude-based infrastructure without fear of a near-term scheduled increase
• The move signals that competition between AI labs is directly benefitting businesses that buy AI services on usage-based terms
• For teams running high-volume Claude workloads such as document processing, customer communications, or knowledge search, the saving compounds at scale

The David and Goliath View

The story behind this pricing reversal is not just about Anthropic. It is about what happens when well-funded AI labs compete aggressively on price. OpenAI slashed its Luna model pricing by 80 per cent. Meta released capable open-source models at no cost under a commercial licence. The result is that Anthropic has opted to make introductory pricing permanent rather than risk losing customers when the September bill arrived.

For lean operators, this is exactly the environment you want to be building in. The cost floor for AI capability keeps dropping, and the gap between what a 15-person business can do with AI and what a 5,000-person enterprise can do with a negotiated enterprise contract is narrowing. The economics of adding AI to customer communications, document processing, or internal knowledge management are more favourable today than they were three months ago.

The practical recommendation is straightforward. If your team paused a Claude project because of the September pricing increase, restart it today. If you never modelled Claude into your AI budget because $3/$15 per million tokens was too expensive, run the numbers now at $2/$10. And if you are already a Claude user, you do not need to do anything. The savings arrive automatically with your next invoice.

Where This Fits in the AI Stack

AI Growth Engine: Permanently lower API costs make it more viable to embed Claude into revenue-generating workflows such as sales communications, proposal drafting, and customer engagement. The reduced cost floor improves the return on investment calculation for outbound and support functions that run at volume.

Employee Amplification Systems: More affordable model access means teams can run more Claude-powered internal tools without hitting budget limits. Research assistance, document summarisation, and meeting preparation tools become cheaper to operate at scale across a team of 20 or 200.

Questions Operators Are Asking

Will the price stay at $2/$10 or could it change again in future? Anthropic has confirmed the $2/$10 rate is the permanent standard price, not the start of a new introductory period. There is no scheduled future increase. That said, AI pricing across the industry is not contractually guaranteed indefinitely, so businesses building on Claude should monitor Anthropic's pricing documentation as they would any vendor contract.

Does this affect Claude Opus 5 or other Claude models? The pricing freeze applies specifically to Claude Sonnet 5. Claude Opus 5, Anthropic's most capable model, is priced separately and at a higher rate. Businesses with mixed workloads should use Sonnet 5 for volume tasks and reserve Opus 5 for complex reasoning or high-stakes outputs where quality is the primary consideration.

How does Claude Sonnet 5 at $2/$10 compare to competitor models? At current market rates, Claude Sonnet 5 is competitively priced against comparable mid-tier models from OpenAI and Google. The exact comparison depends on your workload, as input and output token ratios vary significantly by use case. Running a cost comparison using your actual token volumes is the most reliable approach before making a vendor decision.

Should we switch from another model to Claude Sonnet 5 given the price lock? Not on price alone. Model switching involves integration work and carries performance risk. If Claude Sonnet 5 already meets your performance requirements and the locked pricing improves your economics, a formal evaluation is worth completing. If you are on a model that is performing well, a price advantage on an alternative model is worth modelling but not worth disrupting production systems without a clear quality advantage too.

Is this a good time to start experimenting with Claude for the first time? Yes. The pricing lock removes one of the most common reasons teams delay evaluation. Starting with a low-volume pilot on non-critical workflows carries minimal financial risk and gives your team direct experience with the model before committing to broader integration work.

Citable Summary

What happened: Anthropic confirmed on 10 August 2026 that Claude Sonnet 5 will remain at its introductory price of $2 per million input tokens and $10 per million output tokens permanently, cancelling a 50 per cent increase that was originally scheduled for 1 September 2026.

Why it matters: Any business using or evaluating Claude Sonnet 5 via API has avoided a significant cost increase today, improving the economics of Claude-based workflows and making AI product development more accessible for smaller organisations.

David and Goliath view: Competitive pressure from OpenAI and Meta is driving model pricing down across the industry. Lean operators benefit most when AI capability costs fall, as it reduces the cost advantage that large enterprises hold through negotiated volume contracts.

Offer relevance:

• AI Growth Engine: Lower per-token costs make Claude-powered revenue workflows more financially viable at the volumes a business with 10 to 200 employees would run
• Employee Amplification Systems: More affordable access enables broader internal tool deployment without hitting budget constraints that typically limit smaller teams