Schemas

Error

OpenAI-style error body.

object · required

AnthropicErrorBody

type

string · const · required

Const value: error

object · required

PassthroughObject

Provider-compatible JSON object passed through from the upstream (with `model` set to the catalog slug and `usage.cost` added in USD).

ChatMessage

role

string · enum · required

Enum values:

system

developer

user

assistant

tool

Text, or an array of content parts (text, image_url, …).

name

string

object[]

tool_call_id

string

ChatCompletionRequest

OpenAI Chat Completions request. Only the commonly used fields are listed; other OpenAI fields are passed through.

model

string · required

Model slug from GET /v1/models (a model with mode chat).

ChatMessage[] · minItems: 1 · required

stream

boolean

Stream server-sent events. stream_options.include_usage is forced on.

Default: false

max_tokens

integer

Output limit; lowered automatically when your credit can't cover it.

max_completion_tokens

integer

Output limit (preferred over max_tokens when both are set); lowered automatically when your credit can't cover it.

temperature

number

top_p

number

n

integer

Number of choices; each is reserved and billed.

seed

integer

presence_penalty

number

frequency_penalty

number

object[]

Function tools. Provider-hosted tools (web search, code interpreter, MCP, …) are refused.

parallel_tool_calls

boolean

object

{"type":"json_object"} or a json_schema.

reasoning_effort

string

low, medium or high on models with supports_reasoning.

logprobs

boolean

top_logprobs

integer

object

ChatCompletion

OpenAI `chat.completion` object. `model` is the catalog slug; `usage.cost` is the billed amount in USD.

id

string

object

string

created

integer

model

string

object[]

object

ResponsesRequest

OpenAI Responses request. Other Responses fields are passed through.

model

string · required

required

Text, or a list of input items (input_text, input_image, …). Must not be empty.

instructions

string

stream

boolean

Default: false

max_output_tokens

integer

Output limit; lowered automatically when your credit can't cover it.

object

{"effort": "low"} on reasoning models.

object[]

tool_choice

object

temperature

number

top_p

number

previous_response_id

string

Only ids of responses previously returned to your own account.

store

boolean

MessagesRequest

Anthropic Messages request. Fields not listed are ignored.

model

string · required

Any chat model slug; dated Claude names map to the catalog slug.

object[] · minItems: 1 · required

max_tokens

integer

Required by Anthropic SDKs; lowered automatically when your credit can't cover it.

System prompt; text blocks may carry cache_control (kept for claude* models).

stream

boolean

Default: false

temperature

number

top_p

number

stop_sequences

string[]

object[]

Client tools with name, description, input_schema.

object

{"type": "auto" | "any" | "none" | "tool", "name"?, "disable_parallel_tool_use"?}

object

{"type": "enabled", "budget_tokens": N} or {"type": "adaptive"}; used on reasoning models only.

AnthropicMessage

id

string

type

string · const

Const value: message

role

string · const

Const value: assistant

model

string

object[]

thinking, text and tool_use blocks.

stop_reason

string · enum

Enum values:

end_turn

max_tokens

tool_use

refusal

stop_sequence

null

object

EmbeddingsRequest

model

string · required

An embedding model (mode embedding).

required

One text or a list of texts. Must not be empty.

dimensions

integer

Shorter vectors, where the model supports it.

encoding_format

string

DecisionsRequest

model

string · required

required

The text to judge, or messages with input_text parts.

object[] · required

Questions of type predicate, choice (with choices) or score (with levels). Answers come back in the same order.

DecisionsResponse

model

string

object[]

object

SystemOneRequest

model

string · required

state

required

The text or state to judge.

object · required

Named questions, each {"type": "noul" | "choice" | "score", "instructions": …}.

Model

A catalog model (dashboard format). Prices are nano-USD per 1M tokens.

slug

string · required

The id to send as model.

name

string · required

description

string · required

owned_by

string · required

context_length

integer · required

prompt_price_nano

integer · int64 · required

Input price, nano-USD per 1M tokens.

completion_price_nano

integer · int64 · required

Output price, nano-USD per 1M tokens.

cache_read_price_nano

integer | null · int64 · required

Cached-input price, nano-USD per 1M tokens; null = same as the input price.

created

integer · int64 · required

Unix seconds when the model was added.

mode

string · enum · required

Which endpoints the model works on.

Enum values:

chat

embedding

decision

supports_reasoning

boolean · required

Accepts reasoning_effort (thinking).

supports_vision

boolean · required

Accepts image input.

OpenAIModel

id

string · required

object

string · const · required

Const value: model

created

integer · int64 · required

owned_by

string · required

name

string · required

description

string · required

context_length

integer · required

supports_reasoning

boolean · required

supports_vision

boolean · required

mode

string · enum · required

Enum values:

chat

embedding

decision

object · required

USD per token, as decimal strings with 12 decimals.

OpenAIModelList

object

string · const · required

Const value: list

OpenAIModel[] · required

AnthropicModelList

object[] · required

has_more

boolean · required

first_id

string | null

last_id

string | null

SessionToken

token

string · required

Session token (JWT, valid 7 days). Send as Authorization: Bearer <token>.

VerificationRequired

verification_required

boolean · const · required

Const value:

email

string · email · required

MfaChallenge

token

string · required

The mfa_token for /api/auth/mfa/setup and /api/auth/mfa/verify (43 characters, valid 10 minutes).

setup

boolean · required

True when the user has no 2FA yet and must enrol first (admins).

MfaSetupInfo

secret

string · required

Base32 secret in groups of 4 characters, for manual entry.

otpauth_uri

string · required

otpauth://totp/… URI.

qr_svg

string · required

SVG markup of the QR code for otpauth_uri.

account

string · required

The user's email.

issuer

string · required

The site name.

CodeRequest

code

string · required

6-digit authenticator code (or a recovery code where accepted).

EmailRequest

email

string · email · required

turnstile_token

string | null

Cloudflare Turnstile response, when enabled.

AuthMethods

email_password

boolean · required

Email/password sign-in is enabled.

registration

boolean · required

New accounts can sign up.

google

boolean · required

Google sign-in is configured and enabled.

turnstile_site_key

string | null · required

Cloudflare Turnstile site key when the security check is on; then send turnstile_token on login, register, resend-verification and forgot-password.

Me

id

string · uuid · required

email

string · email · required

name

string · required

is_admin

boolean · required

balance_nano

integer · int64 · required

Credit balance in nano-USD (1 USD = 1,000,000,000). Includes credit currently reserved by in-flight requests as already spent.

mfa_enabled

boolean · required

Two-factor authentication is set up.

has_password

boolean · required

Can sign in with a password (Google-only accounts can't).

google_linked

boolean · required

created_at

string · date-time · required

NotificationPrefs

low_balance_alert

boolean · required

low_balance_threshold_nano

integer | null · int64 · required

Custom threshold in nano-USD; null uses the default.

default_threshold_nano

integer · int64 · required

The default threshold (1 USD = 1000000000).

Branding

site_name

string · required

site_url

string · uri · required

icon_url

string | null · required

Versioned path such as /api/branding/icon?v=3; null = use the built-in logo.

timezone

string · required

IANA time zone for usage dates and charts.

ApiKeyRequest

Settings for create and update. On update the whole object replaces the current settings.

name

string · minLength: 1 · maxLength: 100 · required

credit_limit_nano

integer | null · int64 · min: 0

Spend cap in nano-USD (per budget period, if set). null = no cap beyond the account balance.

budget_period

string | null · enum

Resets the key's spend each period (in the site time zone). Requires credit_limit_nano.

Enum values:

daily

weekly

monthly

allowed_models

array | null · maxItems: 200

Model slugs this key may use. null or empty = all models.

expires_at

string | null · date-time

Must be in the future.

allowed_ips

array | null · maxItems: 50

IP addresses or CIDR ranges (203.0.113.7, 10.0.0.0/8, 2001:db8::/32). null or empty = any address.

rpm_limit

integer | null · min: 1

Requests per minute for this key, up to the account limit. null = the default per-key limit.

max_parallel

integer | null · min: 1 · max: 1000

Requests in progress at once for this key.

redact_secrets

boolean

Mask credentials (API keys, tokens, …) found in prompts before they are sent to the provider.

Default: false

ApiKey

id

string · uuid · required

name

string · required

key_prefix

string · required

Masked key for display, e.g. sk-****abcd.

credit_limit_nano

integer | null · int64 · required

Spend cap in nano-USD.

usage_nano

integer · int64 · required

Spend in nano-USD in the current budget period (or ever, without a period).

allowed_models

array | null · required

expires_at

string | null · date-time · required

allowed_ips

array | null · required

rpm_limit

integer | null · required

max_parallel

integer | null · required

budget_period

string | null · enum · required

Enum values:

daily

weekly

monthly

budget_reset_at

string | null · date-time · required

When the current budget period ends.

redact_secrets

boolean · required

last_used_at

string | null · date-time · required

created_at

string · date-time · required

UsageLog

id

string · uuid · required

model_slug

string · required

provider_slug

string · required

api_key_name

string | null · required

null for playground requests.

prompt_tokens

integer · required

completion_tokens

integer · required

cached_tokens

integer · required

cost_nano

integer · int64 · required

Billed cost in nano-USD (0 for failed requests).

latency_ms

integer · required

ttft_ms

integer | null · required

Time to first token (streamed requests).

tps

number | null · required

Output tokens per second, measured from the first token when streamed.

status_code

integer · required

streamed

boolean · required

source

string · enum · required

Enum values:

api

playground

created_at

string · date-time · required

UsageLogDetail

id

string · uuid · required

upstream_id

string | null · required

The provider's response id.

model_slug

string · required

provider_slug

string · required

api_key_name

string | null · required

source

string · enum · required

Enum values:

api

playground

streamed

boolean · required

status_code

integer · required

finish_reason

string | null · required

prompt_tokens

integer · required

completion_tokens

integer · required

cached_tokens

integer · required

cache_write_tokens

integer · required

reasoning_tokens

integer · required

cost_nano

integer · int64 · required

Billed cost in nano-USD.

cost_breakdown

object | null · required

Cost split in nano-USD.

latency_ms

integer · required

ttft_ms

integer | null · required

tps

number | null · required

object | null · required

Generation parameters and counts (e.g. stream, temperature, max_tokens, messages count, tools count, endpoint), never content.

error_type

string | null · required

e.g. insufficient_credits, model_not_found, model_not_allowed, invalid_request, rate_limited, upstream_unavailable, stream_interrupted, or the provider's error type.

error_message

string | null · required

client_ip

string | null · required

user_agent

string | null · required

started_at

string · date-time · required

created_at

string · date-time · required

When the request finished.

UsageSummary

bucket

string · enum · required

Enum values:

hour

day

object · required

object[] · required

Top 50 models by cost.

DailyUsage

day

string · date · required

requests

integer · int64 · required

tokens

integer · int64 · required

Prompt + completion tokens.

cost_nano

integer · int64 · required

nano-USD.

BillingConfig

enabled

boolean · required

Top-ups can be created.

idr_per_usd

integer · int64 · required

Rupiah per 1 USD of credit.

min_topup_idr

integer · int64 · required

max_topup_idr

integer · int64 · required

payment_method

string · required

Topup

order_id

string · required

amount_idr

integer · int64 · required

Amount paid, whole Rupiah.

credit_nano

integer · int64 · required

Credit added when paid, nano-USD.

status

string · enum · required

Enum values:

pending

completed

failed

expired

payment_link_url

string | null · uri · required

Hosted payment page.

payment_code

string | null · required

QRIS payload to render as a QR code.

paid_at

string | null · date-time · required

created_at

string · date-time · required

TopupPage

Topup[] · required

total

integer · int64 · required

Topup | null · required

Latest pending top-up, whatever page is shown.

completed

integer · int64 · required

Paid top-ups in total.

Redemption

code

string · required

credit_nano

integer · int64 · required

nano-USD.

created_at

string · date-time · required

Affiliate

enabled

boolean · required

When false, no other field is present.

code

string

The user's referral code (8 characters).

link

string · uri

Sign-up link: <site>/register?ref=<code>.

commission_percent

number

commission_months

integer

referee_bonus_percent

number

referrals

integer · int64

Users who signed up through the link.

paying_referrals

integer · int64

Referrals that earned the user a commission.

earned_nano

integer · int64

Total commission earned, nano-USD.

object[]

Latest 50 rewards.