AnthropicTextVision

claude-sonnet-5

claude-sonnet-5 is Anthropic's model at $2.00 input and $10.00 output per million tokens, with a 1049K context window. On a typical 3:1 input-to-output mix that blends to roughly $4.00 per million tokens, placing it in the premium band of the 89 models tracked here — cheaper than 22% of them.

Open cost calculator View provider Open official platform

MODEL OVERVIEW

Context window1049K

Input price$2 / 1M

Output price$10 / 1M

Output speed95 tok/s

Quality score92

Data updatedUpdated Sep 2026

Price verified2026-09-02 · Official

Price noteThe $2 / $10 rate launched as introductory pricing through 2026-08-31; Anthropic has cancelled the planned 1 September increase to $3 / $15, so this is now the standard price

By price, claude-sonnet-5 ranks 61st cheapest of 89 on input and 66th on output. Its quality score of 92 ranks 19th of 89, and on value (quality score ÷ output price) it ranks 61st of 89. Its blended rate sits 30% above the catalog average.

In real billing terms: a support chatbot handling 50,000 turns a month (2,000 input / 500 output tokens each) costs about $450 per month on claude-sonnet-5, while a RAG workload of 20,000 queries (8,000 input / 800 output tokens each) runs about $480. Because output is priced 5.0× input here, long-generation tasks scale cost fastest — cap output length before you optimise anything else.

A 1049K context window holds roughly 1,399 pages of A4 text (at ~750 tokens per page), ranking 8th of 89 here. That is enough to drop in an entire technical manual or several source files at once, which suits long-document summarization and codebase-wide analysis. Output runs at about 95 tok/s (44th fastest of 89), acceptable interactively, though long responses will feel slow.

We track 12 Anthropic models, and claude-sonnet-5's blended rate sits 58% below that provider's average. If your task tolerates a little less headroom, qwen3.5 (Alibaba) blends about 97% cheaper at a quality score of 91, and is worth benchmarking first. Going the other way, deepseek-v4-pro scores 93 on quality for about 75% less.

The closest model in Anthropic's line-up is claude-sonnet-5.5 ($2.00 input / $10.00 output, quality —). Against it, claude-sonnet-5 scores 92 points higher on quality, has a 49K larger context window. When two models sit this close on price, the deciding factor is usually the quality ceiling — run both against your real prompts before picking.

Our read on it

The September increase was cancelled and $2/$10 became the list price — making the hardest mid-tier default to argue against harder still.

The increase scheduled for 1 September never happened

claude-sonnet-5 was $3/$15 in July 2026 and dropped to $2/$10 in August, billed at the time as introductory pricing that expired on 31 August. Anthropic then cancelled the planned return to $3/$15 and made $2/$10 the standard rate. If you raised a cost model by 50% at the end of August, you can put it back.

At $2/$10 the comparison that matters is the previous generation, still listed here: claude-sonnet-4-6 at $3/$15 and quality 90. The newer tier is a third cheaper and two points better, with nothing traded away elsewhere — 1,049K context, 95 tokens/second, a Fast latency rating, and more throughput than Opus 5. Winning on both axes across a generation is rarer than it sounds.

Why it works as the default you stop thinking about

The most expensive part of model selection is rarely tokens. It is the engineering time spent deciding and migrating. Sonnet 5 earns its position by having no obvious weak column: quality 92 covers the overwhelming majority of production traffic, it is fast enough to sit in front of a user, its context is large enough that you can skip building retrieval infrastructure, and its price is mid-range enough that an overrun is a nuisance rather than an incident.

It carries exactly the same use-case tags as Opus 5 — coding, reasoning, long-context — which is the tell that it is a cheaper point on the same curve rather than a cut-down variant. On a RAG workload (8,000 input and 800 output tokens across 20,000 queries a month) the bill splits roughly evenly between input and output, which is far easier to forecast than an output-dominated model.

When not to pick it

Two situations. High-volume work where cost is a hard constraint: qwen3.5 at $0.10/$0.15 and quality 91 has a blended rate under a fiftieth of Sonnet 5's, for one point less. If the workload is large, tolerant of the occasional miss, and does not depend on Anthropic-specific behaviour such as tool-use details, that gap is too big to wave away.

And anything involving audio. Sonnet 5 handles text and vision only, so a speech pipeline needs gpt-5.6-luna or gpt-5.6-sol alongside it. Beyond those two, talking a team out of Sonnet 5 takes a fairly specific reason — which is exactly what makes it a good default.

Estimated monthly cost

Monthly API spend across four common workloads, next to the cheapest model we track.

Workload	Usage assumption	This model	Cheapest model
Support chatbot	2,000 input / 500 output tokens per turn, 50,000 turns per month	$450	$3.00
RAG document Q&A	8,000 input / 800 output tokens per query, 20,000 queries per month	$480	$3.84
Coding assistant	12,000 input / 2,000 output tokens per request, 10,000 requests per month	$440	$3.20
Batch summarization	4,000 input / 400 output tokens per item, 200,000 items per month	$2,400	$19

Derived directly from the rates on this page. Excludes prompt caching, batch discounts and free tiers.

How it compares

The alternatives worth benchmarking alongside it, from the same dataset.

Model	Provider	Blended /1M	Quality	Price delta
claude-sonnet-5	Anthropic	$4.00	92	—
qwen3.5Cheaper option	Alibaba	$0.11	91	-97%
deepseek-v4-proHigher quality	DeepSeek	$0.99	93	-75%
claude-sonnet-5.5Same provider	Anthropic	$4.00	—	+0%
qwen3.7-flashBest value overall	Alibaba	$0.055	82	-99%

Best for

CodingReasoningLong Context

Modalities

TextVision

Where it fits — and where it doesn't

Good fit

Coding: writing, refactoring and debugging across multi-file project context.
Reasoning: multi-step decomposition, planning and tasks that need rigorous inference.
Long documents: a 1049K window ingests a full contract, paper or large codebase in one pass.
Multimodal input: handles images, screenshots and scanned documents directly.

Poor fit

High-frequency batch jobs: at $4.00 blended per million tokens, bills scale quickly at volume.

Frequently asked questions

How much does the claude-sonnet-5 API cost?

claude-sonnet-5 costs $2.00 per million input tokens and $10.00 per million output tokens. On a 3:1 input-to-output mix that blends to about $4.00 per million tokens.

What does claude-sonnet-5 cost per month?

It depends on volume. Using the four worked scenarios on this page: a support chatbot runs about $450/month, RAG document Q&A about $480/month, a coding assistant about $440/month, and batch summarization about $2,400/month. Plug your own token counts into the cost calculator for a tailored figure.

How large is the claude-sonnet-5 context window?

1049K tokens — roughly 1,399 pages of A4 text at ~750 tokens per page, ranking 8th of the 89 models tracked here.

claude-sonnet-5 or qwen3.5 — which is better value?

qwen3.5 blends to about $0.11 per million tokens versus $4.00 for claude-sonnet-5 — roughly 97% cheaper — at a quality score of 91 against 92. If that quality gap does not show up on your task, take the cheaper one; if it does, stay with claude-sonnet-5. Benchmark both on your own data before deciding.

What is claude-sonnet-5 best suited for?

Tagged use cases: coding, reasoning, long-context. Coding: writing, refactoring and debugging across multi-file project context. Less suitable: High-frequency batch jobs: at $4.00 blended per million tokens, bills scale quickly at volume.

How current is this claude-sonnet-5 pricing?

Pricing here was last verified on 2026-09-02 against official documentation. llmprice.app runs a daily collection job, but providers can change rates between runs — confirm against Anthropic's official pricing page before you commit.