message

Overview

Name	message
Type	Resource
Id	anthropic.messages.message

Fields

Name	Datatype	Description
id	string	Unique object identifier.
content	array	The generated content blocks by the model.
model	string	The model that handled the request.
role	string	The conversational role of the generated message.
stop_reason	string	The reason the model stopped generating.
stop_sequence	string	The stop sequence that caused the model to stop, if applicable.
type	string	Object type, which is always "message" for Messages API.
usage	object	Information about token usage and rate limits.

Methods

Name	Accessible by	Required Params
create_message	SELECT	anthropic-version, data__max_tokens, data__messages, data__model

Required Parameters for Message Creation

Parameter	Description
anthropic-version	API version string (e.g., '2023-06-01')
data__max_tokens	Maximum number of tokens to generate
data__messages	Array of message objects with role and content
data__model	Model identifier (e.g., 'claude-3-5-sonnet-20240620')

Model Comparison Table

To help you choose the right model for your needs, we’ve compiled a table comparing the key features and capabilities of each model in the Claude family:

Model	Claude 3.5 Sonnet	Claude 3 Opus	Claude 3 Sonnet	Claude 3 Haiku
Description	Most intelligent model	Powerful model for highly complex tasks	Balance of intelligence and speed	Fastest and most compact model for near-instant responsiveness
Strengths	Highest level of intelligence and capability	Top-level performance, intelligence, fluency, and understanding	Strong utility, balanced for scaled deployments	Quick and accurate targeted performance
Multilingual	Yes	Yes	Yes	Yes
Vision	Yes	Yes	Yes	Yes
Message Batches API	Yes	Yes	No	Yes
API model name	Upgraded version: claude-3-5-sonnet-20241022 Previous version: claude-3-5-sonnet-20240620	claude-3-opus-20240229	claude-3-sonnet-20240229	claude-3-haiku-20240307
Comparative latency	Fast	Moderately fast	Fast	Fastest
Context window	200K	200K	200K	200K
Max output	8192 tokens	4096 tokens	4096 tokens	4096 tokens
Cost (Input / Output per MTok)	$3.00 / $15.00	$15.00 / $75.00	$3.00 / $15.00	$0.25 / $1.25
Training data cut-off	Apr 2024	Aug 2023	Aug 2023	Aug 2023

Query Examples

Basic Message Generation

This example generates a one-sentence summary of StackQL:

SELECT
  model as model,
  role as role,
  stop_reason as stop_reason,
  stop_sequence as stop_sequence,
  JSON_EXTRACT(usage, '$.input_tokens') as input_tokens,
  JSON_EXTRACT(usage, '$.output_tokens') as output_tokens,
  JSON_EXTRACT(json_each.value, '$.text') as content
FROM
  anthropic.messages.message,
  JSON_EACH(content)
WHERE 
  "anthropic-version" = '2023-06-01'
  AND data__model = 'claude-3-5-sonnet-20240620'
  AND data__max_tokens = 1024
  AND data__messages = '[{"role": "user", "content": "one sentence summary of stackql"}]'

Example response:

|----------------------------|-----------|-------------|---------------|--------------|---------------|--------------------------------|
|           model            |   role    | stop_reason | stop_sequence | input_tokens | output_tokens |            content             |
|----------------------------|-----------|-------------|---------------|--------------|---------------|--------------------------------|
| claude-3-5-sonnet-20240620 | assistant | end_turn    | null          |           13 |            46 | StackQL is a SQL-like query    |
|                            |           |             |               |              |               | language and runtime for       |
|                            |           |             |               |              |               | cloud infrastructure, allowing |
|                            |           |             |               |              |               | users to query, analyze, and   |
|                            |           |             |               |              |               | manage cloud resources across  |
|                            |           |             |               |              |               | multiple providers using       |
|                            |           |             |               |              |               | familiar SQL syntax.           |
|----------------------------|-----------|-------------|---------------|--------------|---------------|--------------------------------|

Multi-turn Conversation Example

This example shows how to conduct a multi-turn conversation:

SELECT
  JSON_EXTRACT(json_each.value, '$.text') as content
FROM
  anthropic.messages.message,
  JSON_EACH(content)
WHERE 
  "anthropic-version" = '2023-06-01'
  AND data__model = 'claude-3-5-sonnet-20240620'
  AND data__max_tokens = 1024
  AND data__messages = '[
    {"role": "user", "content": "What is cloud infrastructure?"},
    {"role": "assistant", "content": "Cloud infrastructure refers to the hardware and software components needed for cloud computing, including servers, storage, and networking resources."},
    {"role": "user", "content": "How can StackQL help manage it?"}
  ]'

Using System Prompts

Example showing how to include a system prompt:

SELECT
  JSON_EXTRACT(json_each.value, '$.text') as content
FROM
  anthropic.messages.message,
  JSON_EACH(content)
WHERE 
  "anthropic-version" = '2023-06-01'
  AND data__model = 'claude-3-5-sonnet-20240620'
  AND data__max_tokens = 1024
  AND data__system = 'You are a cloud infrastructure expert focusing on technical details.'
  AND data__messages = '[{"role": "user", "content": "Explain how StackQL helps with cloud resource management"}]'

Usage Notes

Token Usage

• The usage object contains information about token consumption:
  • input_tokens: Number of tokens in the input
  • output_tokens: Number of tokens generated in the response
• Token usage affects billing and rate limits
• Token counts may not exactly match visible content due to internal processing

Response Processing

• Use JSON_EACH to process the content array
• Content is returned as blocks with type and text
• Use JSON_EXTRACT to access nested JSON fields like usage statistics

Best Practices

1. Always specify an appropriate max_tokens value
2. Include the correct anthropic-version header
3. Format message arrays properly as JSON strings
4. Use system prompts for specialized behavior
5. Monitor token usage for cost optimization

Error Handling

Common error scenarios to handle:

• Invalid model specification
• Malformed message JSON
• Token limit exceeded
• Rate limiting
• Authentication failures

Rate Limits

• Token-based rate limiting
• Monitored through the usage object
• Consider implementing backoff strategies for heavy usage

Version Compatibility

• Specify API version using anthropic-version
• Current supported version: '2023-06-01'
• Check Anthropic's documentation for version updates

Advanced Usage Examples

Temperature Control

Adjust response randomness with the temperature parameter:

/* Lower temperature for more focused responses */
SELECT
  JSON_EXTRACT(json_each.value, '$.text') as content
FROM
  anthropic.messages.message,
  JSON_EACH(content)
WHERE 
  "anthropic-version" = '2023-06-01'
  AND data__model = 'claude-3-5-sonnet-20240620'
  AND data__max_tokens = 1024
  AND data__temperature = 1  
  AND data__messages = '[{"role": "user", "content": "List 3 key benefits of using StackQL"}]'

Using Stop Sequences

Configure custom stop sequences to control response length:

SELECT
  JSON_EXTRACT(json_each.value, '$.text') as content,
  stop_reason,
  stop_sequence
FROM
  anthropic.messages.message,
  JSON_EACH(content)
WHERE 
  "anthropic-version" = '2023-06-01'
  AND data__model = 'claude-3-5-sonnet-20240620'
  AND data__max_tokens = 1024
  AND data__stop_sequences = '["END", "STOP"]'
  AND data__messages = '[{"role": "user", "content": "Describe StackQL until you reach a natural conclusion END"}]'

Metadata Tracking

Include metadata for request tracking:

SELECT
  JSON_EXTRACT(json_each.value, '$.text') as content,
  JSON_EXTRACT(usage, '$.input_tokens') as input_tokens,
  JSON_EXTRACT(usage, '$.output_tokens') as output_tokens
FROM
  anthropic.messages.message,
  JSON_EACH(content)
WHERE 
  "anthropic-version" = '2023-06-01'
  AND data__model = 'claude-3-5-sonnet-20240620'
  AND data__max_tokens = 1024
  AND data__metadata = '{"user_id": "u123"}'
  AND data__messages = '[{"role": "user", "content": "What is StackQL?"}]'

Working with Response Data

Content Block Structure

The content field returns an array of blocks with this structure:

{
  "type": "text",
  "text": "The actual content..."
}

Access specific parts using JSON functions:

-- Get just the text content
JSON_EXTRACT(json_each.value, '$.text')

-- Get the content type
JSON_EXTRACT(json_each.value, '$.type')

Usage Statistics Processing

Extract detailed usage information:

SELECT
  JSON_EXTRACT(usage, '$.input_tokens') as input_tokens,
  JSON_EXTRACT(usage, '$.output_tokens') as output_tokens,
  JSON_EXTRACT(usage, '$.cache_creation_input_tokens') as cache_creation_tokens,
  JSON_EXTRACT(usage, '$.cache_read_input_tokens') as cache_read_tokens
FROM
  anthropic.messages.message
WHERE
  -- query conditions

Performance Optimization

Token Usage Optimization

Tips for minimizing token usage:

1. Use precise prompts
2. Set appropriate max_tokens
3. Use system prompts effectively
4. Leverage stop sequences
5. Monitor and analyze usage patterns

Response Processing Optimization

Strategies for efficient response handling:

1. Use appropriate JSON extraction
2. Implement efficient error handling
3. Process responses asynchronously when possible
4. Cache frequently used responses
5. Implement retry logic with exponential backoff