Model introduction
Claude Sonnet 5 is Anthropic's balanced flagship for coding, agents, and enterprise workflows. It targets a strong speed-intelligence mix and enables adaptive thinking by default.
Model capabilities
Specifications come from the cited model material. Catalog tags help identify access features; use the source documentation for exact limits.
Verified specifications
- Official positioning
- Best combination of speed and intelligence
- Context window
- 1,000,000 tokens
- Maximum output
- 128,000 tokens
- Thinking mode
- Adaptive thinking by default
Sources: Anthropic Claude Sonnet 5 update notes
Independent capability evaluation
Only public measurements matched to the exact model ID and reasoning profile are shown; nearby variants are not substituted.
Intelligence index
53
Cohort rank #12 / 186
Output speed
83.9 tok/s
Cohort rank #69 / 186
First-token latency
191.88s
Adaptive reasoning, max effort
A near-frontier score with good decode speed, but max effort adds a very long thinking delay and heavy token use.
Evaluation profile: Adaptive reasoning, max effort. Artificial Analysis · Methodology
Model pricing and access
Prices are read directly from the MoleAPI console catalog and shown by current billing group and context tier.
Your final charge follows the account group shown in the console.
Live pricing source
Pricing table
USD / 1M tokensStandardx1defaultInput $2 · Output $10
Input
$2 / 1M tokens
Output
$10 / 1M tokens
Cache read
$0.2 / 1M tokens
Cache write
$2.5 / 1M tokens
Discountx0.8discountInput $1.6 · Output $8
Input
$1.6 / 1M tokens
Output
$8 / 1M tokens
Cache read
$0.16 / 1M tokens
Cache write
$2 / 1M tokens
Relayx0.3relayInput $0.6 · Output $3
Input
$0.6 / 1M tokens
Output
$3 / 1M tokens
Cache read
$0.06 / 1M tokens
Cache write
$0.75 / 1M tokens
Access protocols
Available billing groups: Standard, Discount, Relay
- anthropic
- POST
- /v1/messages
- openai
- POST
- /v1/chat/completions
Complete model guide
Prepared from vendor sources and independently reviewed. · 2026-07-28
Claude Sonnet 5 is the next-generation model in Anthropic's Sonnet family, positioned to balance speed and intelligence and as the direct upgrade to Claude Sonnet 4.6. According to Anthropic, its primary improvements over Sonnet 4.6 focus on coding and agentic tasks. [1]
Model Positioning and Core Capabilities
Claude Sonnet 5 is designed for coding, tool use, and agentic workloads. Anthropic describes it as the most capable model for agentic tasks in the Sonnet family. The model can be used for task planning and can execute multi-step workflows with tools such as browsers and terminals. [8][9]
Anthropic positions the Sonnet family for real-time agents and high-throughput workloads. Sonnet 5 brings coding, tool use, and everyday work capabilities that previously typically required larger models to the Sonnet tier. [8][9]
Capabilities and Specifications
Claude Sonnet 5 provides a 1 million token context window and a maximum output of 128,000 tokens, with adaptive thinking enabled by default. [1]
| Specification | Details |
|---|---|
| Context window | 1,000,000 tokens; the default and maximum values are the same. [1] |
| Maximum output | 128,000 tokens. [1] |
| Thinking mode | Adaptive thinking is enabled by default and can be disabled with thinking: {type: "disabled"}. [1] |
| Output budget | max_tokens limits the combined total of thinking tokens and final response tokens. [1] |
| Service tier | Priority Tier is not supported. [1] |
Adaptive thinking uses the effort parameter to control reasoning intensity. The previous manual thinking budget configuration, thinking: {type: "enabled", budget_tokens: N}, is no longer supported. [1]
Use Cases
Claude Sonnet 5 focuses on code development and agentic workflows. It can be used for complex code changes, debugging, refactoring, and tasks that require planning and tool coordination. [1][9]
The model is designed for agentic workloads involving browser or terminal tools, and also supports high-frequency everyday knowledge work and automation workflows. [8][9]
The 1 million token context window is suited to tasks that need to retain large amounts of contextual information in a single request. Its maximum output length also supports generating longer code, analyses, or task results. [1]
Important Limitations
Claude Sonnet 5 does not accept non-default temperature, top_p, or top_k parameters. Setting these parameters returns HTTP 400. [1]
Assistant message prefilling is not supported. Related requests return HTTP 400. [1]
This model uses a new tokenizer. Compared with Sonnet 4.6, the same text produces approximately 30% more tokens, depending on the content. This change affects context capacity estimates, output budgets, and token usage for equivalent amounts of text. [1]
Requests involving prohibited or high-risk cybersecurity topics may be refused. These responses still return HTTP 200, but stop_reason is refusal. [1]
Sources
[1] Anthropic Claude Sonnet 5 更新说明 [8] Claude Sonnet [9] Introducing Claude Sonnet 5
Code examples
These examples use MoleAPI's OpenAI-compatible endpoint and run after you replace the API key.
curl https://api.moleapi.com/v1/chat/completions \
-H "Authorization: Bearer $MOLEAPI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-sonnet-5",
"messages": [{"role": "user", "content": "Hello"}]
}'Frequently asked questions
How is claude-sonnet-5 priced?
This page reads prices from the MoleAPI console API and updates with model prices, context tiers, and account groups.
How can I access claude-sonnet-5?
Protocols and endpoints come from the supported_endpoint_types field in the MoleAPI model catalog.
How do I switch an existing project to claude-sonnet-5?
Keep the MoleAPI API address and key, replace the model parameter with the model ID on this page, then check protocol-specific parameter differences.
Where do the claude-sonnet-5 model details come from?
Capabilities and limitations are checked against Anthropic and the other cited pages. Pricing and available protocols come only from the MoleAPI console.