API Reference

• Chat Completions
  • Sample Usage
  • Request Parameters
  • Response (when not using streaming)
  • Response Format (When Streaming is Used)
  • Important Note about max_tokens
  • Web Search
• Deep Research API
  • Sample Usage
  • Retrieving the Result
  • Request Parameters
  • Notes
• Tokenize API
  • Example Usage
  • Request Parameters
  • Response
• Models API
  • Example
  • Index Models
  • Get Model Details
• Object List
  • MessageObject
  • ReasoningObject
  • ResponseFormatObject
  • ResponseFormatJsonSchemaObject
  • StreamOptionObject
  • ChatCompletionToolsObject
  • ChatCompletionToolsFunctionObject
  • ChatCompletionToolChoiceObject
  • ChatCompletionToolChoiceFunctionObject
  • ChoiceObject
  • ChatMessageObject
  • ChatMessageToolCallsObject
  • ChatMessageToolCallsFunctionObject
  • StreamChoiceObject
  • ChatMessageDeltaObject
  • WebSearchOptionsObject
  • UsageObject
  • PlamoUsageDetailObject
  • ModelObject
• Old API Endpoints

Chat Completions

POST https://api.platform.preferredai.jp/v1/chat/completions

This endpoint generates responses from generation AI models based on provided chat messages.

Currently, only the PLaMo Prime model is supported.

Model ID	Context Length	Maximum Output Tokens	Reasoning Capability
plamo-3.0-prime	262,144 tokens	20,480 tokens	Yes
plamo-3.1-prime	262,144 tokens	20,480 tokens	Yes

Sample Usage

For curl, set your PLaMo API key in an environment variable named PLAMO_API_KEY, and for Python libraries, set it in an environment variable named OPENAI_API_KEY.

Request

curl

$ curl \
    -H "Authorization: Bearer ${PLAMO_API_KEY}" \
    -H "Content-Type: application/json" \
    "https://api.platform.preferredai.jp/v1/chat/completions" \
    -d @- << EOF
{
  "messages": [
    {
      "role": "system",
      "content": "You are a travel advisor."
    },
    {
      "role": "user",
      "content": "Please recommend an optimal sightseeing route in Kanazawa from morning until evening."
    }
  ],
  "model": "plamo-3.0-prime"
}
EOF

Python (langchain-openai)

import os
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://api.platform.preferredai.jp/v1",
    model="plamo-3.0-prime",
    streaming=True,
    # other parameters...,
)

messages=[
    {"role": "system", "content": "You are a travel advisor."},
    {"role": "user", "content": "Please recommend an optimal sightseeing route in Kanazawa from morning until evening."},
]

for chunk in llm.stream(messages):
    print(chunk.content, end="", flush=True)

Python (openai)

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.platform.preferredai.jp/v1",
    # other parameters...,
)

completion = client.chat.completions.create(
    model="plamo-3.0-prime",
    messages=[
        {"role": "system", "content": "You are a travel advisor."},
        {"role": "user", "content": "Please recommend an optimal sightseeing route in Kanazawa from morning until evening."},
    ],
    stream=True,
)

for chunk in completion:
    if chunk.choices and chunk.choices[0].delta.content is not None:
        print(chunk.choices[0].delta.content, end="", flush=True)

Response

non-streaming

{
    "id": "chat-eb6da3a371c546c8a8c4629794328c5b",
    "object": "chat.completion",
    "created": 1733220118,
    "model": "plamo-3.0-prime",
    "choices": [
        {
            "index": 0,
            "message": {
                "role": "assistant",
                "content": "The quadratic formula for solving second-degree equations is as follows:...",
                "tool_calls": []
            },
            "logprobs": null,
            "finish_reason": "stop",
            "stop_reason": null
        }
    ],
    "usage": {
        "prompt_tokens": 169,
        "total_tokens": 394,
        "completion_tokens": 225
    },
    "prompt_logprobs": null
}

streaming

data: {"id":"chat-e05656fa4721494a93d54410881f00c8","object":"chat.completion.chunk","created":1734345308,"model":"plamo-3.0-prime","choices":[{"index":0,"delta":{"role":"assistant"},"logprobs":null,"finish_reason":null}],"usage":{"prompt_tokens":143,"total_tokens":143,"completion_tokens":0}}

data: {"id":"chat-e05656fa4721494a93d54410881f00c8","object":"chat.completion.chunk","created":1734345308,"model":"plamo-3.0-prime","choices":[{"index":0,"delta":{"content":"PL"},"logprobs":null,"finish_reason":null}],"usage":{"prompt_tokens":143,"total_tokens":144,"completion_tokens":1}}

data: {"id":"chat-e05656fa4721494a93d54410881f00c8","object":"chat.completion.chunk","created":1734345308,"model":"plamo-3.0-prime","choices":[{"index":0,"delta":{"content":"a"},"logprobs":null,"finish_reason":null}],"usage":{"prompt_tokens":143,"total_tokens":145,"completion_tokens":2}}

data: {"id":"chat-e05656fa4721494a93d54410881f00c8","object":"chat.completion.chunk","created":1734345308,"model":"plamo-3.0-prime","choices":[{"index":0,"delta":{"content":"Mo"},"logprobs":null,"finish_reason":null}],"usage":{"prompt_tokens":143,"total_tokens":146,"completion_tokens":3}}

data: {"id":"chat-e05656fa4721494a93d54410881f00c8","object":"chat.completion.chunk","created":1734345308,"model":"plamo-3.0-prime","choices":[{"index":0,"delta":{"content":""},"logprobs":null,"finish_reason":"stop","stop_reason":null}],"usage":{"prompt_tokens":143,"total_tokens":148,"completion_tokens":5}}

data: {"id":"chat-e05656fa4721494a93d54410881f00c8","object":"chat.completion.chunk","created":1734345308,"model":"plamo-3.0-prime","choices":[],"usage":{"prompt_tokens":143,"total_tokens":148,"completion_tokens":5}}

data: [DONE]

For the /chat/completions endpoint, models supporting reasoning capabilities support both reasoning and reasoning_content. Non-streaming responses include reasoning and reasoning_content within each choice.message, while streaming responses (when stream: true) incrementally include reasoning and reasoning_content in each chunk's choice.delta.

Request Parameters

Please include the following parameters in the request body in JSON format.

Parameter	Type	Required	Default Value	Description
model	string	required		Specifies the ID of the model to use.
messages	array of MessageObject	required		Sets the list of chat messages. Each message must include a role and content.
frequency_penalty	float	optional	0.0	Controls the frequency of token repetition. Valid values range from -2.0 to 2.0.
max_tokens	integer or null	optional	20480	Stops generation once the specified number of tokens have been produced. Depending on the resulting content, generation may terminate earlier. Cannot be set to exceed the model's maximum output token limit.
n	integer	optional	1	Specifies the number of times to generate content in a single request. Valid values are 1 or 2.
presence_penalty	float	optional	0.0	Applies a penalty for tokens that have already appeared in the output. Valid values range from -2.0 to 2.0.
reasoning	ReasoningObject	optional	null	Allows configuration of reasoning-related settings.
reasoning_effort	none or medium	optional	none	Specifies the level of reasoning to perform. Only none (no reasoning) or medium (standard amount of reasoning) are supported. If effort is specified in reasoning, this setting will be ignored.
response_format	ResponseFormatObject or null	optional	null	Specifies the response output format. Set { "type": "json_schema", "json_schema": {...} } to generate a JSON response that follows the supplied JSON Schema.
seed	integer or null	optional	null	Specifies the random seed for token generation. Depending on internal state, different results may occur even when using the same seed value.
safety_identifier	string or null	optional	null	An identifier for user identification. This may be used by the PLaMo API side to detect anomalies or other issues. Since it may be persisted on the server, do not include any personal information or sensitive data.
stop	array of strings, strings, or null	optional	null	Stops generation when any of the specified words are encountered.
stream	boolean	optional	false	Set to 'true' when using streaming responses.
stream_options	StreamOptionObject or null	optional	null	Specify options when enabling streaming. See StreamOptionObject below for details.
temperature	float	optional	0.3	Specifies the sampling temperature. Valid values range from 0 to 2.0.
tools	array of ChatCompletionToolsObject	optional	null	Defines the functions to be used for Function Calling.
tool_choice	ChatCompletionToolChoiceObject or "none" or "auto" or "required"	optional	null	Specifies the function to be invoked.
top_p	float	optional	1.0	Controls how far into the token generation results to consider. Valid values range from 0 to 1.0.
web_search_options	WebSearchOptionsObject or null	optional	null	Searches the web and answers with citations based on the search results. See Web Search for details.

Response (when not using streaming)

Field	Type	Description
id	string	A unique ID assigned for each message.
choices	array of ChoiceObject	Returns chat-style generation results.
created	integer	Returns the generated UNIX timestamp.
model	string	Returns the model ID.
object	string	Always returns 'chat.completion'.
usage	UsageObject	Returns an object containing usage statistics.
plamo_usage_details	array of PlamoUsageDetailObject or null	PLaMo's usage details breakdown (web search counts, tokens, etc.).

Response Format (When Streaming is Used)

The response is returned using server-sent events. The response will conclude with an event: [DONE].

Field	Type	Description
id	string	A unique ID assigned for each message.
choices	array of StreamChoiceObject	Returns chat-style generation results.
created	integer	Unix timestamp of when the generation was completed.
model	string	Returns the model ID.
object	string	Always returns 'chat.completion.chunk'.
usage	UsageObject or null	Returns an object containing usage statistics.
plamo_usage_details	array of PlamoUsageDetailObject or null	PLaMo's usage details breakdown. Included in the final usage chunk.

Important Note about max_tokens

Currently, the API will output an error when the sum of input tokens + max_tokens exceeds the model's context length. Even if the actual generated output contains fewer tokens than the context length when rendered, the API will still produce an error during the input phase.

Specifically, if max_tokens is not explicitly set, it defaults to 20480, which means an error will occur even if the input length is below the context limit. In such cases, please explicitly specify max_tokens to ensure that input tokens + max_tokens does not exceed the model's context length.

Web Search

When web_search_options is specified, the model searches the web as needed and answers with citations based on the search results. The supported model is plamo-3.1-prime.

Request

curl

$ curl \
    -H "Authorization: Bearer ${PLAMO_API_KEY}" \
    -H "Content-Type: application/json" \
    "https://api.platform.preferredai.jp/v1/chat/completions" \
    -d @- << EOF
{
  "model": "plamo-3.1-prime",
  "messages": [
    {"role": "user", "content": "Where is the head office of Preferred Networks?"}
  ],
  "web_search_options": {
    "search_context_size": "low",
    "user_location": {
      "type": "approximate",
      "approximate": {"country": "JP", "city": "Tokyo", "timezone": "Asia/Tokyo"}
    }
  }
}
EOF

Response

non-streaming

{
    "id": "chatcmpl-621c7cbd-a821-48b1-90b9-70c142934562",
    "object": "chat.completion",
    "model": "plamo-3.1-prime",
    "choices": [
        {
            "index": 0,
            "message": {
                "role": "assistant",
                "content": "The head office of Preferred Networks is in Otemachi Building, 1-6-1 Otemachi, Chiyoda-ku, Tokyo. ([preferred.jp](https://www.preferred.jp/en/company))",
                "annotations": [
                    {
                        "type": "url_citation",
                        "url_citation": {
                            "url": "https://www.preferred.jp/ja/company",
                            "title": "Company - Preferred Networks, Inc.",
                            "start_index": 52,
                            "end_index": 93
                        }
                    }
                ],
                "tool_calls": []
            },
            "finish_reason": "stop"
        }
    ],
    "usage": {
        "prompt_tokens": 20796,
        "completion_tokens": 383,
        "total_tokens": 21179
    },
    "plamo_usage_details": [
        {"model": "plamo-3.1-prime", "usage_type": "main_response", "input_tokens": 20796, "output_tokens": 383, "token_consumed": 21179},
        {"model": "web_search_basic", "usage_type": "web_search", "input_tokens": 1, "output_tokens": 0, "token_consumed": 1}
    ]
}

• The answer text carries a source link in the form ([domain](URL)) after each passage quoted from a page.
• message.annotations lists the sources the answer cites as url_citation. start_index / end_index point to the cited span in the text; a citation of search results points to a zero-width position at the end of the sentence.
• The number of searches is the token_consumed of the plamo_usage_details entry whose usage_type is web_search. Charges are based on the number of tokens and the number of searches.
• reasoning_effort and reasoning (including summary) can be specified. With stream, the answer arrives after the search.
• When response_format specifies a JSON Schema, the answer follows that format and carries no citations.
• When the output limit is reached, finish_reason is length.
• The following parameters cannot be specified (a 400 error is returned): temperature, top_p, presence_penalty, frequency_penalty, n, seed, stop, logprobs, tools, tool_choice.
• Web search cannot be used with messages that contain images.

Deep Research API

POST https://api.platform.preferredai.jp/v1/responses

Searches the web repeatedly on a question and writes a report with citations. It runs as an asynchronous job in the same shape as the OpenAI Responses API with background: true: creating it immediately returns a Response object with status: "queued". Generation takes several minutes to tens of minutes, so retrieve the result with the returned id.

Model ID	Description
plamo-3.1-prime-deep-research	The model for Deep Research. Any other model ID does not run Deep Research.

Sample Usage

Request

curl

$ curl \
    -H "Authorization: Bearer ${PLAMO_API_KEY}" \
    -H "Content-Type: application/json" \
    "https://api.platform.preferredai.jp/v1/responses" \
    -d @- << EOF
{
  "model": "plamo-3.1-prime-deep-research",
  "input": "Research recent trends in generative AI regulation in Japan",
  "background": true
}
EOF

Response

right after creation

{
    "id": "resp_3f9a1c...",
    "object": "response",
    "created_at": 1791272415.0,
    "status": "queued",
    "background": true,
    "model": "plamo-3.1-prime-deep-research",
    "output": [],
    "usage": null,
    "tool_usage": null,
    "error": null,
    "tools": [{"type": "web_search"}],
    "max_tool_calls": 100,
    "incomplete_details": null
}

after completion

{
    "id": "resp_3f9a1c...",
    "object": "response",
    "status": "completed",
    "output": [
        {
            "type": "web_search_call",
            "id": "ws_...",
            "status": "completed",
            "action": {"type": "search", "query": "generative AI regulation Japan", "queries": ["generative AI regulation Japan"]}
        },
        {
            "type": "reasoning",
            "id": "rs_...",
            "summary": [{"type": "summary_text", "text": "..."}]
        },
        {
            "type": "message",
            "id": "msg_...",
            "role": "assistant",
            "status": "completed",
            "content": [
                {
                    "type": "output_text",
                    "text": "# Generative AI Regulation in Japan\n\n...",
                    "annotations": [
                        {"type": "url_citation", "start_index": 120, "end_index": 180, "url": "https://...", "title": "..."}
                    ]
                }
            ]
        }
    ],
    "usage": {"input_tokens": 182340, "output_tokens": 21530, "total_tokens": 203870},
    "tool_usage": {"web_search": {"num_requests": 18}}
}

Retrieving the Result

Endpoint	Description
GET /v1/responses/{id}	Returns the job's current Response object. status moves from queued to in_progress and ends as completed, incomplete, failed, or cancelled.
GET /v1/responses/{id}?stream=true	Returns the progress as server-sent events, starting from the events that have already occurred. If the connection drops, pass the sequence_number of the last event you received as starting_after to resume from there.
POST /v1/responses/{id}/cancel	Cancels the job. A cancelled job ends with status: "cancelled".

The stream carries the same response.* events as the OpenAI Responses API. The main events are:

• response.queued / response.in_progress / response.completed
• response.output_item.added / response.web_search_call.searching
• response.reasoning_summary_text.delta
• response.output_text.delta / response.output_text.annotation.added

Results are kept for 24 hours after creation. A job created by another tenant cannot be retrieved (404 is returned).

Request Parameters

Please include the following parameters in the request body in JSON format. Any parameter not listed here returns an error.

Parameter	Type	Required	Default Value	Description
model	string	required		Specify plamo-3.1-prime-deep-research.
input	string or array of MessageObject	required		What to research. For an array, the last user message is the question, and assistant messages after it are used as the research plan. system messages are not used.
background	boolean	required		Specify true. Any other value returns an error.
stream	boolean	optional	false	true is not supported at creation. Get the progress with GET /v1/responses/{id}?stream=true.
tools	array	optional	null	Restricts the source domains in the form [{"type": "web_search", "filters": {"allowed_domains": [...], "blocked_domains": [...]}}]. No tool other than web_search and no option other than filters can be specified.
max_tool_calls	integer	optional	100	The maximum total number of searches, page opens, and in-page finds in one job. Valid values range from 1 to 100. Once the limit is reached, the remaining chapters are left out.
include	array of string	optional	null	Specify ["web_search_call.action.sources"] to include the search result URLs in action.sources of each web_search_call.
metadata	object	optional	null	Any string keys and values.
store	boolean	optional	null	Accepted for compatibility with the OpenAI Responses API. Regardless of the value, results are kept for 24 hours after creation.

Notes

• A job runs for at most about 25 minutes. Past that, it ends with status: "failed" and error.code: "timeout".
• If part of the research (such as writing a chapter heading) fails, that part is skipped, the report is written to the end, and the job ends with status: "incomplete". What is missing is written in the summary of a reasoning item.
• Charges are based on the tokens used to write the report and the number of searches in tool_usage.web_search.num_requests.

Tokenize API

POST https://api.platform.preferredai.jp/v1/tokenize

Converts a given string into tokens and returns both the token sequence and the total number of tokens.

Note that in Chat Completion, the text will be converted to a chat format, so various prompts will be automatically inserted. Therefore, the token count returned here may not strictly match the actual number of tokens generated in the final text output.

Also, since PLaMo uses the same Tokenizer as this publicly available model, you can integrate similar functionality directly into your own system rather than using the API.

Example Usage

Request

curl

$ curl \
    -H "Authorization: Bearer ${PLAMO_API_KEY}" \
    -H "Content-Type: application/json" \
    "https://api.platform.preferredai.jp/v1/tokenize" \
    -d '{"prompt": "Hello", "model": "plamo-3.0-prime"}'

Response

curl

{
    "count": 4,
    "max_model_len": 16384,
    "tokens": [
        1,
        1,
        35913,
        37607
    ]
}

Request Parameters

Specify the following parameters in JSON format within the request body.

Parameter	Type	Required	Default Value	Description
model	string	yes		Specify the ID of the model to be used.
prompt	string	Either messages or prompt must be specified		Specify the string for which you want to calculate tokens.
messages	array of MessageObject	Either messages or prompt must be specified		Specify the strings for which you want to calculate tokens. The same format as chat messages used in the Chat Completions endpoint can be used.
add_special_tokens	boolean	no	true	Specify whether to add special tokens like BOS.
add_generation_prompt	boolean	no	true	Specify whether to apply internal generation prompts during tokenization. This option is only valid when messages are specified.

Response

Field	Type	Description
count	integer	Total number of tokens.
max_model_len	integer	Maximum number of tokens the model can process.
tokens	array of integer	The actual token sequence.

Models API

Retrieves data including available models.

Example

Request

For curl, set your PLaMo API key in the PLAMO_API_KEY environment variable, and for Python libraries, set it in the OPENAI_API_KEY environment variable.

curl

$ curl \
    -H "Authorization: Bearer ${PLAMO_API_KEY}" \
    -H "Content-Type: application/json" \
    "https://api.platform.preferredai.jp/v1/models"

Python(openai)

from openai import OpenAI
client = OpenAI(
    base_url="https://api.platform.preferredai.jp/v1",
)

models = client.models.list()
print(models)

model_name = models.data[0].id
model = client.models.retrieve(model_name)
print(model)

Response

curl

{
    "data": [
        {
            "id": "plamo-3.0-prime",
            "created": 1732978800,
            "object": "model",
            "owned_by": "system",
            "max_model_len": 262144
        }
    ],
    "object": "list"
}

Index Models

GET https://api.platform.preferredai.jp/v1/models

Retrieves a list of available models.

Response

Field	Type	Description
object	string	Always contains the value list.
data	array of ModelObject	Returns an array of model objects

Get Model Details

GET https://api.platform.preferredai.jp/v1/models/{model}

Fetches detailed information about a specified model.

Path Parameters

Parameter	Type	Required	Description
model	string	yes	Specifies the ID of the model to retrieve.

Response

Returns a ModelObject.

Object List

MessageObject

Represents a chat message, primarily consisting of System, User, and Assistant components.

Parameter	Type	Required	Default Value	Description
content	string	yes		Specifies the content of the message.
role	string	yes		Indicates the role of the message; must be one of 'system', 'user', or 'assistant'.
name	string	no		A name used for further identification when the same role is referenced.
reasoning	string	no		Sets the reasoning provided in the response.
reasoning_content	string	no		Sets the reasoning_content provided in the response.

ReasoningObject

Specifies parameters related to reasoning.

Parameter	Type	Required	Default Value	Description
summary	string	no	null	Specifies whether to generate a summary of the reasoning content. Can be one of 'auto', 'concise', or 'detailed'. If not specified, summary generation is disabled. Note that tokens used for summary generation will be counted as token usage.
effort	none or medium	no	none	Allows selection of reasoning intensity. Supported options are none (no reasoning) or medium (standard amount of reasoning). Takes precedence over reasoning_effort.

ResponseFormatObject

Specifies the response output format.

Parameter	Type	Required	Default Value	Description
type	string	yes		Must be set to json_schema.
json_schema	ResponseFormatJsonSchemaObject	yes		Defines the JSON structure that the model should generate, expressed as JSON Schema.

ResponseFormatJsonSchemaObject

Specifies the JSON Schema configuration used for Structured Outputs.

Parameter	Type	Required	Default Value	Description
name	string	yes		The name of the response format.
description	string or null	no	null	A description of the response format.
schema	object	yes		The JSON Schema object that the response should follow. For the JSON Schema specification, see JSON Schema.
strict	boolean	no	false	When set to true, the model will generate responses that strictly follow the structure defined by schema. Supported JSON Schema features are limited.

StreamOptionObject

Specifies options when using streaming responses.

Parameter	Type	Required	Default Value	Description
include_usage	boolean	no		Specifies whether to return usage statistics just before the streaming response completes.

ChatCompletionToolsObject

Defines the functions to be used with the Function Calling feature.

Field	Type	Description
type	string	Must be set to 'function'.
function	ChatCompletionsToolsFunctionObject	Defines the actual function to be used with Function Calling.

ChatCompletionToolsFunctionObject

Defines the actual function implementation.

Field	Type	Description
name	string	The name of the function.
description	string	A description of the function.
parameters	string	Definition of arguments that can be passed to the function, expressed in JSON Schema. For details on JSON Schema, see: https://json-schema.org/understanding-json-schema

ChatCompletionToolChoiceObject

Specifies which function to use.

Field	Type	Description
type	string	Must be set to 'function'.
function	ChatCompletionToolChoiceFunctionObject	Specifies the function to use.

ChatCompletionToolChoiceFunctionObject

Specifies the function details.

Field	Type	Description
name	string	Specifies the name of the function specified in the 'tools' parameter.

ChoiceObject

Represents the result of generating a response in chat format.

Field	Type	Description
index	integer	Returns the index of the object when multiple ChoiceObjects are available.
finish_reason	'stop' or 'length'	Returns 'stop' if the specified word was encountered or if generation naturally concluded; returns 'length' if the maximum token limit was exceeded.
logprobs	null	Currently not supported.
message	ChatMessageObject	Returns the content of the generated message.
stop_reason	integer, string, or null	Currently not supported.

ChatMessageObject

Represents the content of a generated message.

Field	Type	Description
content	string	Returns the message content.
reasoning	string or null	Returns the model's reasoning process. This field may only be present for models supporting reasoning.
reasoning_content	string or null	Contains the same content as reasoning.
annotations	array or null	With web search, returns the sources the answer cites as url_citation. See Web Search for details.
role	string	Returns the role of the message, which can be one of 'system', 'user', or 'assistant'.
tool_calls	ChatMessageToolCallsObject	Information about Function Calling generated by the model.

ChatMessageToolCallsObject

Indicates the result of a function call.

Field	Type	Description
id	string	A unique identifier.
type	string	Must be 'function'.
function	ChatMessageToolCallsFunctionObject	Information about the Function called via Function Calling.

ChatMessageToolCallsFunctionObject

Information about the called Function.

Field	Type	Description
name	string	Indicates the name of the Function.
arguments	string	Function arguments specified in JSON format.

StreamChoiceObject

Represents the result of generated chat messages in a stream format.

Field	Type	Description
index	integer	Returns the index of the object when multiple ChoiceObject instances exist.
finish_reason	'stop' or 'length'	Returns stop if the specified word is encountered or generation ends naturally, and length if the maximum generated token count is exceeded.
logprobs	null	Currently not supported.
delta	ChatMessageDeltaObject	Returns the content of the additionally generated message.
stop_reason	integer or string or null	Currently not supported.

ChatMessageDeltaObject

Represents the content of a message in a stream format.

Field	Type	Description
content	string	Returns the content of the message. The content field may be absent in cases such as when a role assignment is emitted.
reasoning	string or null	Returns the model's reasoning process. This field may be present only for models that support reasoning.
reasoning_content	string or null	Contains the same content as reasoning.
annotations	array or null	With web search, returns the cited sources as url_citation. In streaming, it is included in the final chunk.
role	string	Returns the role of the message, which is one of system, user, or assistant. This field is set only when a new role is assigned.

WebSearchOptionsObject

Specifies the web search settings.

Parameter	Type	Required	Default Value	Description
search_context_size	low or medium or high	optional	medium	How much of the answer's context goes to search results: low uses 1/4, medium 1/2, and high all of it.
user_location	object or null	optional	null	The user's approximate location, in the form {"type": "approximate", "approximate": {"country": "JP", "city": "Tokyo", "region": "Tokyo", "timezone": "Asia/Tokyo"}}. Each field of approximate can be omitted.
filters	object or null	optional	null	Restricts the source domains in the form {"allowed_domains": ["example.com"]}. 1 to 100 domains can be specified.

UsageObject

Indicates token generation results and related information.

Field	Type	Description
completion_tokens	integer	Returns the number of generated tokens.

PlamoUsageDetailObject

Indicates PLaMo-specific breakdown per model and usage type.

Field	Type	Description
model	string	The model or pseudo-model name used (e.g. plamo-3.1-prime, web_search_basic).
usage_type	string	The usage type (e.g. main_response, web_search).
input_tokens	integer	Number of input tokens (or web search request count).
output_tokens	integer	Number of output tokens.
token_consumed	integer	Total consumed token count (or billing unit).

ModelObject

Indicates information about a model.

Field	Type	Description
id	string	A string that identifies the model. It can be used as an API parameter.
created	integer	The time when the model was created.
object	string	The type of this object. Currently, this field is always model.
owned_by	string	The owner of the model. Currently, this field is always system.
max_model_len	integer	The maximum number of tokens a single request can use, counting the prompt and the generated text together.

Old API Endpoints

The API endpoints under https://platform.preferredai.jp/api/completion/v1 have been migrated. The new endpoint is https://api.platform.preferredai.jp/v1.

The old API endpoints remain available, but they are deprecated. Please use the new API endpoints.