# Claude Sonnet 5: Pricing, Benchmarks, and Cost

> Claude Sonnet 5 costs $2/$10 per million tokens after Anthropic cancelled the rise to $3/$15. What changed versus Sonnet 4.6, and the cost per task.

Source: https://capitalandcompute.net/blog/claude-sonnet-5-pricing-benchmarks/
Published: 2026-07-01
Updated: 2026-09-29

Key takeaways

* Claude Sonnet 5 shipped on June 30, 2026 and is now the default Claude model, replacing Sonnet 4.6.
* Pricing is $2 per million input tokens and $10 per million output tokens, a third under the $3 / $15 that Sonnet 4.6 charged.
* That rate launched as introductory through August 31, 2026, but Anthropic cancelled the scheduled increase and made it standard. Nothing reverts.
* On agentic coding it closes most of the gap to Opus 4.8 while costing 40% less per output token.

Update, September 29, 2026: Sonnet 5.5 replaced Sonnet 5

Claude Sonnet 5.5 became Anthropic’s mid-tier model on September 28, 2026, at exactly the Sonnet 5 rate card of $2 and $10 per million tokens. Sonnet 5 stays on the API, and it is the model Sonnet 5.5 falls back to for higher-risk cybersecurity requests. What changed and what it costs per task is in [Claude Sonnet 5.5 pricing and benchmarks](/blog/claude-sonnet-5-5-pricing-benchmarks/).

Anthropic released Claude Sonnet 5 on June 30, 2026, and made it the default model on the Free and Pro plans the same day. For anyone budgeting an AI workload, the headline is a price cut that took three months to become permanent: $2 per million input tokens and $10 per million output, a third under the $3 / $15 that Sonnet 4.6 charged. It arrived as a time-boxed discount and stayed. Alongside it came a capability jump that moves Sonnet into territory that used to require the Opus tier.

Here is what Sonnet 5 costs, what changed against Sonnet 4.6 and [Opus 4.8](/blog/claude-fable-5-vs-opus-4-8-cost-per-task/), and whether any of it changes the cost per task. If you are deciding between tiers, see [Claude Opus 5.5 vs Sonnet 5](/blog/claude-opus-vs-sonnet/), since Anthropic now recommends starting most workloads on Opus 5.5.

## How much does Claude Sonnet 5 cost?

API pricing is $2 per million input tokens and $10 per million output tokens, per [Anthropic’s pricing documentation](https://platform.claude.com/docs/en/about-claude/pricing). That is a third below the $3 / $15 Sonnet 4.6 charged.

This rate arrived as introductory pricing through August 31, 2026, announced in the [Claude Sonnet 5 launch post](https://www.anthropic.com/news/claude-sonnet-5), with a scheduled rise to $3 / $15 on September 1\. **The increase did not happen.** Anthropic’s pricing page now records that $2 / $10 is the standard price and that the scheduled increase will not occur. This post originally advised budgeting against $3 / $15; that advice was correct when written and is no longer.

| Model                      | Input / Mtok | Output / Mtok | Cache read / Mtok | Position                    |
| -------------------------- | ------------ | ------------- | ----------------- | --------------------------- |
| Claude Sonnet 5            | $2.00        | $10.00        | $0.20             | mid tier until Sep 28, 2026 |
| Claude Sonnet 4.6 (legacy) | $3.00        | $15.00        | $0.30             | superseded, still available |
| Claude Opus 5.5            | $4.00        | $20.00        | $0.20             | current flagship            |
| Claude Opus 4.8            | $5.00        | $25.00        | $0.50             | superseded, still available |
| Claude Haiku 4.5           | $1.00        | $5.00         | $0.10             | cheapest tier               |

Rates re-verified September 23, 2026.

A discount that became the price

Treating the $2 / $10 as temporary was the right call for two months and the wrong one afterwards. The general lesson still holds: an introductory rate with a stated expiry is not a durable budgeting basis. The specific lesson is the other half of it, which is that a vendor can cancel its own increase, so a rate needs re-reading rather than assuming either direction.

The prompt cache read is $0.20 per million tokens, 10% of input, which matters because coding agents lean heavily on cached context. For the dated, sourced rate card across every current model, the [AI model tracker](/ai-models/) keeps them current.

## What changed from Sonnet 4.6

The specs are a clean upgrade at the same price. Sonnet 5 carries a 1M-token context window, 128k maximum output, adaptive thinking, and a January 2026 knowledge cutoff, per [Anthropic’s models overview](https://platform.claude.com/docs/en/about-claude/models/overview).

| Spec              | Claude Sonnet 5 |
| ----------------- | --------------- |
| Context window    | 1M tokens       |
| Max output        | 128k tokens     |
| Adaptive thinking | Yes             |
| Knowledge cutoff  | January 2026    |
| API id            | claude-sonnet-5 |
| Released          | June 30, 2026   |

On the benchmarks Anthropic published with the launch, Sonnet 5 is a straight improvement over Sonnet 4.6 across reasoning, coding, and computer use. Anthropic presents the head-to-head as a chart in the launch post; the numbers below are that table as transcribed by [DataCamp](https://www.datacamp.com/blog/claude-sonnet-5), with the two readable primary anchors (OSWorld-Verified and Humanity’s Last Exam for Sonnet 4.6) matching the launch post directly.

__Claude Sonnet 4.6 versus Sonnet 5 across four launch benchmarks__
| Item                              | Sonnet 5 | Sonnet 4.6 |
| --------------------------------- | -------- | ---------- |
| SWE-bench Pro (coding)            | 63.2%    | 58.1%      |
| Terminal-Bench 2.1                | 80.4%    | 67%        |
| OSWorld-Verified (computer use)   | 81.2%    | 78.5%      |
| Humanity's Last Exam (with tools) | 57.4%    | 46.8%      |

Sonnet 4.6 to Sonnet 5 on four of Anthropic's launch benchmarks. The gold dot is Sonnet 4.6, the slate dot is Sonnet 5; the line is the gain. Terminal-Bench 2.1 moves the most, up 13.4 points. All measured at the rate card in force at launch.Source: [Anthropic, Claude Sonnet 5 launch benchmarks, as transcribed by DataCamp](https://www.anthropic.com/news/claude-sonnet-5)

## Sonnet 5 versus Opus 4.8: the value question

The reason this release matters for a budget is the gap it closes to Opus. On SWE-bench Pro, Anthropic’s harder agentic-coding benchmark, Sonnet 5 scores 63.2% against Opus 4.8 at 69.2%, per the launch benchmarks. Sonnet 5 buys most of the way to Opus-grade coding at $15 output versus Opus at $25, a 40% saving on the expensive side of the meter.

__Claude models: SWE-bench Pro score against output price__
| Item       | SWE-bench Pro score (%) | Output $ / Mtok |
| ---------- | ----------------------- | --------------- |
| Sonnet 5   | 63.2%                   | $15             |
| Sonnet 4.6 | 58.1%                   | $15             |
| Opus 4.8   | 69.2%                   | $25             |

Agentic coding capability against output price. Sonnet 5 (highlighted) sits close to Opus 4.8 on SWE-bench Pro while costing $15 per million output tokens against Opus at $25\. Sonnet 4.6 shares Sonnet 5's price but trails it on the benchmark.Source: [Anthropic, Claude Sonnet 5 launch benchmarks (SWE-bench Pro), as transcribed by DataCamp; standard output rates from Anthropic models overview](https://www.anthropic.com/news/claude-sonnet-5)

This is the standing trade in the Claude line: Opus for the hardest, highest-autonomy work where a few extra points of reliability pay for themselves, Sonnet for the daily-driver volume where cost per finished task decides the bill. Sonnet 5 widens the set of work where the cheaper tier is the correct call. To cost a specific matchup, the [model comparison tool](/ai-model-comparison/) prices Sonnet 5 against Opus 4.8, DeepSeek V4, and the rest on the same task.

## Does it change the cost per task?

Yes, by about a third, and permanently. A modeled multi-file coding task ran about $1.46 on Sonnet 4.6 at $3 / $15\. At the $2 / $10 Sonnet 5 now bills, the same token profile costs about a third less. The [cost-per-task math](/blog/claude-code-pricing-cost-per-task/) still describes the method; only the rate it is applied to has moved, and the [cost-per-task calculator](/ai-coding-cost-calculator/) carries the current figure.

Token efficiency still matters more than the sticker. A stronger model that finishes in fewer output tokens, or with fewer retries, costs less per task even at the same rate. That is the lens worth budgeting on, the same way it applied to the [GPT-5.6 pricing question](/blog/did-openai-cut-ai-prices-gpt-5-6/).

## Should you switch?

For most Claude workloads the switch already happened: Sonnet 5 is the default, and at the same standard price with better benchmarks there is no reason to pin Sonnet 4.6 unless a specific evaluation depends on the older snapshot. Three rules cover the decision:

1. **Coming from Sonnet 4.6, just move.** Same price, higher scores, same 1M context. Pin 4.6 only if you need a frozen model for reproducibility.
2. **Coming from Opus 4.8, test Sonnet 5 on your hardest real task.** If it clears your bar, you drop from $25 to $10 output. If it does not, the top tier is now [Claude Opus 5.5](/blog/claude-opus-5-5-pricing-benchmarks/) at $4 / $20, which is 2x Sonnet rather than the 2.5x Opus 4.8 represented.
3. **Stop treating the rate as temporary.** The scheduled September 2026 increase was cancelled and $2 / $10 is the standard price.

Sonnet 5 turned out to be a rare kind of release: more capability at a third less, once the discount stopped being one. The number to watch is still dollars per finished task on your own workload rather than the rate on the card, but here the card itself moved in the right direction.

## Frequently asked questions

How much does Claude Sonnet 5 cost?

API pricing is $2 per million input tokens and $10 per million output tokens, a third below the $3 / $15 that Sonnet 4.6 charged. That rate launched as introductory pricing through August 31, 2026, but Anthropic cancelled the scheduled increase and made $2 / $10 the standard price.

Is Claude Sonnet 5 more expensive than Sonnet 4.6?

No, it is cheaper. Sonnet 5 bills $2 input and $10 output per million tokens against Sonnet 4.6 at $3 and $15, a third less for higher benchmark scores.

What is the context window and output limit of Claude Sonnet 5?

Sonnet 5 has a 1M-token context window and a 128k-token maximum output, with adaptive thinking and a January 2026 knowledge cutoff, per Anthropic's models overview. Its API id is claude-sonnet-5.

How does Claude Sonnet 5 compare to Opus 4.8?

On SWE-bench Pro, Anthropic's agentic-coding benchmark, Sonnet 5 scores 63.2% against Opus 4.8 at 69.2%, while costing $15 per million output tokens versus Opus at $25\. Sonnet 5 buys most of Opus-grade coding for 40% less on output.

Should I switch from Sonnet 4.6 to Sonnet 5?

For most workloads, yes. Sonnet 5 is the default, scores higher on every launch benchmark, and costs the same at the standard rate. Keep Sonnet 4.6 pinned only when you need a frozen snapshot for reproducibility.

## Sources

* Anthropic. (2026). _Introducing Claude Sonnet 5_ \[vendor launch post; release date, positioning, introductory pricing, launch benchmarks\]. Verified 2026-07-01\. [anthropic.com/news/claude-sonnet-5](https://www.anthropic.com/news/claude-sonnet-5)
* Anthropic. (2026). _Models overview_ \[Claude Platform docs; 1M context, 128k output, adaptive thinking, Jan 2026 cutoff, `claude-sonnet-5` API id\]. Verified 2026-07-01\. [platform.claude.com/docs/en/about-claude/models/overview](https://platform.claude.com/docs/en/about-claude/models/overview)
* DataCamp. (2026). _Claude Sonnet 5: Features, Benchmarks, and Pricing_ \[secondary; transcription of Anthropic’s launch benchmark table for SWE-bench Pro, Terminal-Bench 2.1, OSWorld-Verified, and Humanity’s Last Exam\]. Verified 2026-07-01\. [datacamp.com/blog/claude-sonnet-5](https://www.datacamp.com/blog/claude-sonnet-5)
* Anthropic. (2026). _Pricing_ \[Claude Platform docs; $2 / $10 confirmed as the standard rate, with the scheduled September 1, 2026 increase to $3 / $15 recorded as cancelled\]. Verified 2026-09-23\. [platform.claude.com/docs/en/about-claude/pricing](https://platform.claude.com/docs/en/about-claude/pricing)