Skip to main content
POST
/
api
/
dataframer
/
generate
JavaScript
import Dataframer from 'dataframer';

const client = new Dataframer({
  apiKey: 'My API Key',
});

const generate = await client.dataframer.generate.create({
  generation_model: 'anthropic/claude-sonnet-4-5',
  number_of_samples: 1,
  spec_id: '182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e',
});

console.log(generate.run_id);
{
  "task_id": "<string>",
  "status": "ACCEPTED",
  "run_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a"
}

Authorizations

​
Authorization
string
header
required

API Key authentication. Format: "Bearer YOUR_API_KEY"

Body

application/json
​
spec_id
string<uuid>
required

ID of the spec to use for generation

​
number_of_samples
integer
required

Number of samples to generate

Required range: 1 <= x <= 20000
​
generation_model
enum<string>
required

AI model to use for generation

Available options:
anthropic/claude-sonnet-4-5,
anthropic/claude-sonnet-4-5-thinking,
anthropic/claude-haiku-4-5,
deepseek-ai/DeepSeek-V3.1,
moonshotai/Kimi-K2-Instruct,
openai/gpt-oss-120b,
deepseek-ai/DeepSeek-R1-0528-tput,
Qwen/Qwen2.5-72B-Instruct-Turbo
​
spec_version_id
string<uuid>

Specific version ID to use (optional, defaults to latest version)

​
evaluation_model
enum<string>

AI model for evaluation (short samples only)

Available options:
anthropic/claude-sonnet-4-5,
anthropic/claude-sonnet-4-5-thinking,
anthropic/claude-haiku-4-5,
deepseek-ai/DeepSeek-V3.1,
moonshotai/Kimi-K2-Instruct,
openai/gpt-oss-120b,
deepseek-ai/DeepSeek-R1-0528-tput,
Qwen/Qwen2.5-72B-Instruct-Turbo
​
outline_model
enum<string>

AI model for outline generation (long samples only)

Available options:
anthropic/claude-sonnet-4-5,
anthropic/claude-sonnet-4-5-thinking,
anthropic/claude-haiku-4-5,
deepseek-ai/DeepSeek-V3.1,
moonshotai/Kimi-K2-Instruct,
openai/gpt-oss-120b,
deepseek-ai/DeepSeek-R1-0528-tput,
Qwen/Qwen2.5-72B-Instruct-Turbo
​
revision_model
enum<string>

AI model for revisions (long samples only)

Available options:
anthropic/claude-sonnet-4-5,
anthropic/claude-sonnet-4-5-thinking,
anthropic/claude-haiku-4-5,
deepseek-ai/DeepSeek-V3.1,
moonshotai/Kimi-K2-Instruct,
openai/gpt-oss-120b,
deepseek-ai/DeepSeek-R1-0528-tput,
Qwen/Qwen2.5-72B-Instruct-Turbo
​
enable_revisions
boolean
default:false

Enable revision cycles

​
sample_type
enum<string>
default:short

Type of samples to generate

Available options:
short,
long
​
max_iterations
integer

Max feedback iterations (short samples only)

Required range: 0 <= x <= 20
​
staged_generation
boolean

Use staged generation approach (short samples only)

​
use_historical_feedback
boolean

Use historical feedback (short samples only)

​
num_examples_in_prompt
integer

Number of examples to include in prompt (short samples only)

Required range: 1 <= x <= 50
​
max_revision_cycles
integer

Max revision cycles (long samples only)

Required range: 1 <= x <= 5
​
generation_thinking_budget
integer

Thinking budget for generation model (tokens)

Required range: x >= 1024
​
evaluation_thinking_budget
integer

Thinking budget for evaluation model (tokens, short samples)

Required range: x >= 1024
​
outline_thinking_budget
integer

Thinking budget for outline model (tokens, long samples)

Required range: x >= 1024
​
revision_thinking_budget
integer

Thinking budget for revision model (tokens, long samples)

Required range: x >= 1024
​
seed_shuffling_level
enum<string>
default:sample

Seed shuffling level for long samples. Controls trade-off between prompt caching efficiency and data diversity.

Available options:
none,
sample,
field,
prompt
​
sql_validation_level
enum<string>
default:syntax+schema+execute

SQL validation level for long samples with SQL content

Available options:
syntax,
syntax+schema,
syntax+schema+execute
​
max_examples_in_prompt
integer

Maximum number of seed examples to include in prompts (long samples only). If not set, all seeds are used (subject to token limits).

Required range: x >= 1

Response

Generation started successfully

​
task_id
string
required

Task ID for tracking generation progress

​
status
enum<string>
required

Initial status of the generation task

Available options:
ACCEPTED,
PENDING,
RUNNING
​
run_id
string<uuid>
required

Run ID for retrieving results