Built for developers too

One key.
Two clean APIs.

Automate the platform your agency runs, or give a client direct model access. Every key has explicit scopes and optional spend caps from day one.

Get an API key Read the reference

OpenAPI 3.1 AI-readable docs Familiar model clients

quickstart.py Python

from openai import OpenAI

client = OpenAI(
  base_url="https://models.bytepilot.ai/v1",
  api_key="bp_live_..."
)

response = client.chat.completions.create(
  model="bytepilot",
  messages=[{
    "role": "user",
    "content": "Hello from my app"
  }]
)

print(response.choices[0].message)

CompatibleUsage appears in the same BytePilot balance

Automate your accountPlatform APIRead and control your account. Explore → Build AI productsModel APICall AI models directly, on your account. Explore →

API reference

Everything needed to ship.

Both APIs authenticate with keys created in the BytePilot dashboard. Give each key only the scopes it needs, then add a spend cap if the integration should never cross a fixed boundary.

Read and control your account.

Platform API

Your agents, clients, calls, transcripts, credit balance and usage reports — the data behind the dashboard, in JSON. This is what you automate against: pull last month's usage per client and raise your own invoices, sync clients from your CRM, pause an agent from your own admin panel.

Base URLhttps://bytepilot.ai/api/v1

Every endpoint below is this prefix plus its own path.

AuthenticationSend a business API key as X-API-Key (or Authorization: Bearer). Create keys on your dashboard's Developers page.

Pull this month's usage, rolled up per client

curl https://bytepilot.ai/api/v1/usage \
  -H 'X-API-Key: bp_live_your_key_here'

Endpoints

28 public methods

Open any method for parameters and examples

GET https://bytepilot.ai/api/v1/agents List your agents ⌄

Every agent on your account with its status, template, assigned phone number, and client grouping (external_ref round-trips your own CRM/accounting ids).

Required scopeagents

Response · 200

{
    "agents": [
        {
            "agent_id": 12,
            "name": "Sophie",
            "template": "generic_reception",
            "status": "ACTIVE",
            "phone_number": "+441256222333",
            "client": {
                "client_id": 3,
                "name": "Harrison & Co",
                "external_ref": "CRM-1042"
            },
            "created_at": "2026-07-28 09:00:00"
        }
    ]
}

POST https://bytepilot.ai/api/v1/agents/:id/status Pause or resume an agent ⌄

Paused agents answer with a polite unavailable message. The change applies from the next call.

Required scopeagents

Parameters

idpath · required

Agent id

Request

{
    "status": "PAUSED"
}

Response · 200

{
    "agent_id": 12,
    "status": "PAUSED"
}

GET https://bytepilot.ai/api/v1/clients List your clients ⌄

The client groupings your agents roll up to for billing.

Required scopeagents

Response · 200

{
    "clients": [
        {
            "client_id": 3,
            "name": "Harrison & Co",
            "external_ref": "CRM-1042",
            "created_at": "2026-07-01 10:00:00"
        }
    ]
}

POST https://bytepilot.ai/api/v1/clients Create a client (idempotent by name) ⌄

Creates a client grouping, or returns the existing one with that name (updating external_ref if supplied) — safe to call from sync jobs.

Required scopeagents

Request

{
    "name": "Harrison & Co",
    "external_ref": "CRM-1042"
}

Response · 200

{
    "client_id": 3,
    "name": "Harrison & Co",
    "external_ref": "CRM-1042",
    "created": true
}

GET https://bytepilot.ai/api/v1/visibility/targets List tracked websites ⌄

Every website you track for AI visibility, with its latest score and the client it belongs to. AI visibility measures how often leading AI assistants (ChatGPT, Claude, Gemini, Perplexity) recommend a business when people ask the buying questions its customers actually ask.

Required scopevisibility

Response · 200

{
    "targets": [
        {
            "target_id": 5,
            "url": "https://millerplumbing.co.uk",
            "domain": "millerplumbing.co.uk",
            "display_name": "Miller Plumbing",
            "category": "emergency plumber",
            "location": "Leeds",
            "client": {
                "client_id": 3,
                "name": "Miller Plumbing",
                "external_ref": "CRM-1042"
            },
            "latest_report_id": 41,
            "latest_score": 43,
            "created_at": "2026-08-01 09:00:00"
        }
    ]
}

POST https://bytepilot.ai/api/v1/visibility/targets Track a website (idempotent by domain) ⌄

Reads the website once to derive the business name, category, location and the set of buyer-style questions every report will measure against — so scores stay comparable over time. Returns the existing target if the domain is already tracked. Creating a target is free; running reports costs credit.

Required scopevisibility

Request

{
    "url": "https://millerplumbing.co.uk",
    "client_name": "Miller Plumbing"
}

Response · 200

{
    "target_id": 5,
    "domain": "millerplumbing.co.uk",
    "display_name": "Miller Plumbing",
    "category": "emergency plumber",
    "location": "Leeds",
    "created": true
}

POST https://bytepilot.ai/api/v1/visibility/targets/:id/run Run a visibility check ⌄

Queues a full report: every measured assistant is asked every buyer question, the website gets technical AI-readiness checks, and recommendations are drafted. Reports take a few minutes and you are only charged (flat rate per report, shown as price_pence) when one completes — a failed run is free. Optional send_to / copy_to email the finished report automatically under your branding with the PDF attached, exactly like the dashboard send.

Required scopevisibility

Parameters

idpath · required

Target id

Request

{
    "send_to": "owner@millerplumbing.co.uk",
    "copy_to": "you@youragency.co.uk"
}

Response · 200

{
    "report_id": 42,
    "status": "QUEUED",
    "price_pence": 200,
    "note": "Reports take a few minutes. Poll GET /api/v1/visibility/reports/{report_id}. You are only charged when it completes."
}

GET https://bytepilot.ai/api/v1/visibility/targets/:id/reports List a website's reports ⌄

Report history for one tracked website, newest first (up to 50) — the score trend an agency charts for its client.

Required scopevisibility

Parameters

idpath · required

Target id

Response · 200

{
    "reports": [
        {
            "report_id": 41,
            "status": "COMPLETE",
            "score": 43,
            "charged_pence": 200,
            "scheduled": false,
            "created_at": "2026-08-18 08:00:00",
            "completed_at": "2026-08-18 08:07:00"
        }
    ]
}

GET https://bytepilot.ai/api/v1/visibility/reports/:id Get a report ⌄

The full report: overall 0–100 score, per-assistant breakdown, who the assistants recommend instead, the website's AI-readiness checks and the drafted recommendations. Completed reports also include agent_prompt — a ready-to-paste brief for an AI coding agent working on the measured website, containing the baseline, the buyer questions being lost and a prioritised task list; hand it to your own tooling or to the client's developer. Pass include_answers=1 for the question-by-question evidence. Poll this after queuing a run — status moves QUEUED → RUNNING → COMPLETE (or FAILED, uncharged).

Required scopevisibility

Parameters

idpath · required

Report id

include_answersquery

Set 1 to include every assistant answer excerpt

Response · 200

{
    "report_id": 41,
    "target_id": 5,
    "status": "COMPLETE",
    "score": 43,
    "assistants": [
        {
            "assistant": "ChatGPT",
            "available": true,
            "questions_answered": 10,
            "times_recommended": 4,
            "score": 40
        }
    ],
    "competitors": [
        {
            "name": "Aqua Flow Plumbers",
            "mentions": 5
        }
    ],
    "site_checks": {
        "schema_org": false,
        "llms_txt": false,
        "blocked_crawlers": [
            "GPTBot"
        ]
    },
    "recommendations": {
        "summary": "Strong on emergency questions, invisible on boiler servicing.",
        "gaps": [
            "Blocked GPTBot means ChatGPT cannot read the site"
        ],
        "actions": [
            "Unblock GPTBot in robots.txt"
        ]
    },
    "agent_prompt": "You are working on the website for Miller Plumbing (https://millerplumbing.co.uk)\u2026\nTASKS, in priority order:\n1. robots.txt currently blocks GPTBot\u2026",
    "charged_pence": 200,
    "scheduled": false,
    "created_at": "2026-08-18 08:00:00",
    "completed_at": "2026-08-18 08:07:00"
}

GET https://bytepilot.ai/api/v1/visibility/reports/:id/pdf Download a report as PDF ⌄

The client-ready PDF, carrying YOUR branding (name, logo, colour from your dashboard branding page) and nothing of ours — the same document the dashboard sends to clients. Binary application/pdf response.

Required scopevisibility

Parameters

idpath · required

Report id (must be COMPLETE)

Response · 200

{
    "note": "Binary PDF body, Content-Type: application/pdf"
}

GET https://bytepilot.ai/api/v1/leads/campaigns List lead campaigns ⌄

Your lead-intelligence campaigns, newest first. A campaign finds businesses matching an industry and location, researches each one's website and web presence, scores how strong a prospect it is for the services you sell, and drafts the outreach — you are billed a flat rate per prospect scored, never for discovery or failed research.

Required scopeleads

Response · 200

{
    "campaigns": [
        {
            "campaign_id": 7,
            "label": "Dentists in Manchester",
            "industry": "dentists",
            "location": "Manchester",
            "services": [
                "Websites",
                "AI receptionist"
            ],
            "status": "COMPLETE",
            "target": 25,
            "discovered": 25,
            "scored": 23,
            "failed": 2,
            "created_at": "2026-08-27 09:00:00",
            "completed_at": "2026-08-27 09:41:00"
        }
    ]
}

POST https://bytepilot.ai/api/v1/leads/campaigns Start a lead campaign ⌄

Queues discovery and research. Your balance must cover the campaign's maximum cost (count × per-prospect rate) up front, but charges land per prospect actually scored. Optional send_to / copy_to email the finished white-label prospect report (your branding, PDF attached) when the campaign completes. exclude_previous (default true) skips businesses already found for the SAME client (campaigns with no client form their own scope) — so a monthly re-run stays fresh, while the same lead can still surface for a different client it also suits. Optional seller_context (a sentence or two on what the seller is and offers) tailors the scoring, pitches and drafts to what is genuinely on offer — the dashboard derives it automatically from the seller's website. Better still, pass profile_id (a seller profile created in the dashboard): the campaign then also carries the seller's confirmed proof points, ideal customer, sign-off name and booking link, and drafts come out ready to send. Profiles are read fresh at scoring time, so profile edits apply to every later run.

Required scopeleads

Request

{
    "industry": "dentists",
    "location": "Manchester",
    "count": 25,
    "services": [
        "Websites",
        "AI receptionist"
    ],
    "client_name": "Smile Group",
    "seller_context": "Smile Group: a dental marketing agency offering websites and AI reception for practices.",
    "send_to": "owner@smilegroup.co.uk",
    "copy_to": "you@youragency.co.uk"
}

Response · 200

{
    "campaign_id": 7,
    "status": "QUEUED",
    "price_per_prospect_pence": 25,
    "max_cost_pence": 625,
    "note": "Campaigns take several minutes. Poll GET /api/v1/leads/campaigns/{campaign_id}. You are only charged per prospect scored."
}

GET https://bytepilot.ai/api/v1/leads/campaigns/:id Get a campaign ⌄

Progress and totals for one campaign. Poll this after starting one — status moves QUEUED → DISCOVERING → RESEARCHING → COMPLETE (or FAILED, uncharged).

Required scopeleads

Parameters

idpath · required

Campaign id

Response · 200

{
    "campaign_id": 7,
    "label": "Dentists in Manchester",
    "industry": "dentists",
    "location": "Manchester",
    "services": [
        "Websites",
        "AI receptionist"
    ],
    "status": "RESEARCHING",
    "target": 25,
    "discovered": 25,
    "scored": 11,
    "failed": 1,
    "charged_pence": 275,
    "created_at": "2026-08-27 09:00:00",
    "completed_at": null
}

GET https://bytepilot.ai/api/v1/leads/campaigns/:id/prospects List a campaign's scored prospects ⌄

Scored prospects, best first, with per-service fit scores and the recommended pitch. Filter with min_score to feed only strong prospects into your own tooling.

Required scopeleads

Parameters

idpath · required

Campaign id

min_scorequery

Only prospects scoring at least this (0-100)

Response · 200

{
    "prospects": [
        {
            "prospect_id": 91,
            "name": "Smile Dental",
            "website": "https://smiledental.example",
            "domain": "smiledental.example",
            "phone": "0161 000 0000",
            "lead_score": 91,
            "fit_scores": {
                "Websites": 41,
                "AI receptionist": 96
            },
            "pitch": "Lead with the AI receptionist \u2014 no out-of-hours cover and no online booking."
        }
    ]
}

GET https://bytepilot.ai/api/v1/leads/prospects/:id Get a prospect in full ⌄

Everything the research produced: the observed website signals (deterministic), best-effort external notes, per-service fit scores, the personalised angle, and the outreach drafts (cold email and call script) ready to send from YOUR OWN tools — the platform never sends outreach itself.

Required scopeleads

Parameters

idpath · required

Prospect id

Response · 200

{
    "prospect_id": 91,
    "campaign_id": 7,
    "name": "Smile Dental",
    "website": "https://smiledental.example",
    "domain": "smiledental.example",
    "contact_email": "hello@smiledental.example",
    "phone": "0161 000 0000",
    "status": "SCORED",
    "lead_score": 91,
    "fit_scores": {
        "Websites": 41,
        "AI receptionist": 96
    },
    "signals": [
        {
            "signal": "Online booking",
            "level": "warn",
            "detail": "No visible way to book online \u2014 customers must call during opening hours."
        }
    ],
    "external_notes": {
        "reviews": "around 23 Google reviews",
        "competitors": [
            "Brighter Smiles"
        ]
    },
    "pitch": "Lead with the AI receptionist.",
    "angle": "Their site asks patients to phone for appointments, with no way to book or ask questions outside opening hours.",
    "email_subject": "Out-of-hours enquiries at Smile Dental",
    "email_draft": "Hi, I was looking at smiledental.example and noticed\u2026",
    "call_script": "OPENER: \u2026",
    "charged_pence": 25
}

GET https://bytepilot.ai/api/v1/leads/campaigns/:id/pdf Download a campaign report as PDF ⌄

The client-ready prospect report, carrying YOUR branding and nothing of ours — the same document the dashboard emails to clients. Binary application/pdf response.

Required scopeleads

Parameters

idpath · required

Campaign id (must be COMPLETE)

Response · 200

{
    "note": "Binary PDF body, Content-Type: application/pdf"
}

GET https://bytepilot.ai/api/v1/calls List calls in a date range ⌄

Filterable by agent_id and client_id; defaults to the current month. Latest 200.

Required scopecalls

Parameters

fromquery

YYYY-MM-DD (default: first of this month)

toquery

YYYY-MM-DD (default: today)

agent_idquery

Filter to one agent

client_idquery

Filter to one client

Response · 200

{
    "from": "2026-07-01",
    "to": "2026-07-28",
    "calls": [
        {
            "call_id": 88,
            "agent_id": 12,
            "channel_id": 15,
            "client_id": 3,
            "caller_number": "+447712345678",
            "started_at": "2026-07-28 14:03:11",
            "duration_seconds": 151,
            "status": "COMPLETED",
            "outcome": "message_taken",
            "summary": "Quote request",
            "charged_pence": 26
        }
    ]
}

GET https://bytepilot.ai/api/v1/calls/:id One call with transcript and message ⌄

The full record: timings, outcome, the structured message taken (if any) and the conversation transcript (subject to your transcript retention window).

Required scopecalls

Parameters

idpath · required

Call id

Response · 200

{
    "call_id": 88,
    "agent_id": 12,
    "channel_id": 15,
    "client_id": 3,
    "caller_number": "+447712345678",
    "to_number": "+441256222333",
    "started_at": "2026-07-28 14:03:11",
    "ended_at": "2026-07-28 14:05:42",
    "duration_seconds": 151,
    "status": "COMPLETED",
    "outcome": "message_taken",
    "summary": "Quote request",
    "message": {
        "caller_name": "John Peters",
        "reason": "Quote request",
        "details": "Ltd company, ~40 invoices/month"
    },
    "transcript": [
        "[0:01] Agent: Hi, this is Sophie\u2026"
    ],
    "charged_pence": 26,
    "rate_ppu_used": 10
}

GET https://bytepilot.ai/api/v1/balance Current credit balance ⌄

Balance in pence, billing mode, and account status.

Required scopebilling

Response · 200

{
    "balance_pence": 892,
    "billing_mode": "CREDITS",
    "account_status": "ACTIVE"
}

GET https://bytepilot.ai/api/v1/ledger Credit ledger entries in a date range ⌄

Every balance movement: grants, top-ups, usage debits (with the API key that incurred them, where applicable), adjustments. Latest 500 in range.

Required scopebilling

Parameters

fromquery

YYYY-MM-DD (default: first of this month)

toquery

YYYY-MM-DD (default: today)

Response · 200

{
    "from": "2026-07-01",
    "to": "2026-07-28",
    "entries": [
        {
            "entry_id": 14,
            "type": "USAGE",
            "amount_pence": -13,
            "service": "voice",
            "reference": "CA3f84\u2026",
            "balance_after_pence": 892,
            "api_key_id": null,
            "created_at": "2026-07-28 10:46:20"
        }
    ]
}

GET https://bytepilot.ai/api/v1/usage Usage report: units + cost per client, agent and channel ⌄

The endpoint your billing automation calls monthly to invoice your clients: usage rolled up channel → agent → client with your external_ref on every row, filtered by date range and optionally client_id / agent_id / channel_id.

Required scopebilling

Parameters

fromquery

YYYY-MM-DD (default: first of this month)

toquery

YYYY-MM-DD (default: today)

client_idquery

Filter to one client

agent_idquery

Filter to one agent

channel_idquery

Filter to one channel

Response · 200

{
    "from": "2026-07-01",
    "to": "2026-07-28",
    "usage": [
        {
            "service": "voice",
            "client_id": 3,
            "client_name": "Harrison & Co",
            "external_ref": "CRM-1042",
            "agent_id": 12,
            "agent_name": "Sophie",
            "channel_id": 15,
            "calls": 42,
            "seconds": 6510,
            "charged_pence": 1110
        }
    ]
}

POST https://bytepilot.ai/api/v1/outbound/calls Ring someone now ⌄

Places a call immediately — the endpoint your CRM hits when a customer asks to be called back. The permission that makes the call lawful is recorded from the `consent` object in the same request, so there is no way to schedule a call without saying why you may make it. Every call, from here or the scheduler, passes the same gate at the moment of dialling: do-not-call list, permission on file, calling hours, and the per-number and per-agent limits. A refused call answers **409 with the reason in plain words** — the request was fine, we simply must not make that call — and is still recorded so it shows in your history.

Required scopeoutbound

Request

{
    "agent_id": 12,
    "to_number": "+447712345678",
    "reason": "Quote request from the website",
    "consent": {
        "basis": "requested",
        "source": "Website callback form, 29 July",
        "tps_attested": false
    }
}

Response · 200

{
    "outbound_id": 84,
    "status": "placed",
    "reason": ""
}

GET https://bytepilot.ai/api/v1/outbound/calls Outbound call history, including refused ones ⌄

Every attempt in the range with what became of it. Refused attempts carry `block_reason`, which is the answer to "why didn't it ring them" — the commonest question this feature produces.

Required scopeoutbound

Parameters

fromquery

YYYY-MM-DD (default: first of this month)

toquery

YYYY-MM-DD (default: today)

Response · 200

{
    "from": "2026-07-01",
    "to": "2026-07-29",
    "calls": [
        {
            "outbound_id": 84,
            "agent_id": 12,
            "agent_name": "Ellie",
            "client_id": 3,
            "schedule_id": null,
            "to_number": "+447712345678",
            "purpose": "callback",
            "status": "COMPLETED",
            "block_reason": null,
            "call_sid": "CA9f2\u2026",
            "created_at": "2026-07-29 10:14:02"
        }
    ]
}

POST https://bytepilot.ai/api/v1/outbound/schedules Schedule recurring calls ⌄

Sets up a repeating call — a morning check-in, a weekly follow-up. `time_of_day` is in the RECIPIENT's timezone, so nine in the morning stays nine in the morning when the clocks change. Cadence is once, daily, weekdays or weekly (`day_of_week` 1–7, Monday first). As with an immediate call, the permission is recorded from `consent` in the same request, and re-checked every time the schedule fires rather than trusted from when it was created.

Required scopeoutbound

Request

{
    "agent_id": 12,
    "to_number": "+447712345678",
    "label": "Morning check-in with Doris",
    "cadence": "daily",
    "time_of_day": "09:00",
    "timezone": "Europe/London",
    "consent": {
        "basis": "requested",
        "source": "Arranged with her daughter, 12 July"
    }
}

Response · 200

{
    "schedule_id": 7,
    "next_run_at": "2026-07-30 08:00:00",
    "status": "ACTIVE"
}

GET https://bytepilot.ai/api/v1/outbound/schedules List your schedules ⌄

Every schedule on the account with its next call time in UTC.

Required scopeoutbound

Response · 200

{
    "schedules": [
        {
            "schedule_id": 7,
            "agent_id": 12,
            "label": "Morning check-in with Doris",
            "to_number": "+447712345678",
            "cadence": "daily",
            "time_of_day": "09:00",
            "day_of_week": null,
            "timezone": "Europe/London",
            "next_run_at": "2026-07-30 08:00:00",
            "status": "ACTIVE"
        }
    ]
}

POST https://bytepilot.ai/api/v1/outbound/schedules/:id/cancel Stop a schedule ⌄

No further calls are placed. Attempts already made stay in your history.

Required scopeoutbound

Parameters

idpath · required

Schedule id

Response · 200

{
    "schedule_id": 7,
    "status": "CANCELLED"
}

POST https://bytepilot.ai/api/v1/outbound/do-not-call Add a number to the do-not-call list ⌄

Nothing overrides this: it is checked at the moment of dialling, and recording fresh permission afterwards does not undo it. `scope` is `account` (every client you run — the default) or `client` with a `client_id` (that one business only). Your agents add entries here themselves whenever somebody asks not to be called again, so you rarely need this.

Required scopeoutbound

Request

{
    "phone_number": "+447712345678",
    "scope": "account",
    "note": "Asked by email"
}

Response · 200

{
    "phone_number": "+447712345678",
    "scope": "account",
    "status": "suppressed"
}

GET https://bytepilot.ai/api/v1/outbound/do-not-call List do-not-call entries ⌄

Who your agents will never ring, and why each one is on the list.

Required scopeoutbound

Response · 200

{
    "entries": [
        {
            "phone_number": "+447712345678",
            "scope": "client",
            "client_id": 3,
            "reason": "asked_on_call",
            "note": "Asked during a call",
            "created_at": "2026-07-29 10:22:41"
        }
    ]
}

Call AI models directly, on your account.

Model API

A drop-in replacement for the OpenAI API. Change two lines in any existing OpenAI-compatible app — the base URL and the key — and it runs on your BytePilot balance instead. Same request and response shapes, streaming included; we forward your request untouched and hand back exactly what came out.

It exists so the technical end of your book has somewhere to go. A client who wants to build their own thing gets a key from you, spends against your balance, and appears as a line on your usage report — so you can bill them for it at whatever margin you set. Give each client their own key, cap it, and read its spend separately.

Billing is a multiple of what the request actually costs us, not a per-token rate — model prices move monthly and a fixed rate would be wrong within weeks. Fractions of a penny accumulate rather than rounding up, so a thousand tiny requests cost what a thousand tiny requests should.

Base URLhttps://models.bytepilot.ai/v1

Include the /v1 — this is the exact string that goes in an OpenAI client's base_url, and the client appends /chat/completions itself.

AuthenticationThe SAME business API key, with the models scope. Send it as Authorization: Bearer (what OpenAI clients do by default) or X-API-Key.

Any OpenAI client — point it at us and go

from openai import OpenAI

client = OpenAI(
    base_url="https://models.bytepilot.ai/v1",
    api_key="bp_live_your_key_here",
)

reply = client.chat.completions.create(
    model="google/gemini-3.1-flash-lite",
    messages=[{"role": "user", "content": "Summarise this enquiry in one line."}],
)
print(reply.choices[0].message.content)

Or straight over HTTP

curl https://models.bytepilot.ai/v1/chat/completions \
  -H 'Authorization: Bearer bp_live_your_key_here' \
  -H 'Content-Type: application/json' \
  -d '{
        "model": "google/gemini-3.1-flash-lite",
        "messages": [{"role": "user", "content": "Hello"}]
      }'

Endpoints

2 public methods

Open any method for parameters and examples

POST https://models.bytepilot.ai/v1/chat/completions Create a chat completion ⌄

The OpenAI `/v1/chat/completions` contract, unchanged: same request body, same response body, same SSE frames when you set `stream: true`. We do not rewrite your prompt, substitute your model or inject anything — an API that surprises you is worse than one that does less. Every request is attributed to the key that made it, so per-client cost reporting comes free, and a key over its spend cap gets a 402 rather than a surprise bill.

Required scopemodels

Request

{
    "model": "google/gemini-3.1-flash-lite",
    "messages": [
        {
            "role": "user",
            "content": "Summarise this enquiry in one line."
        }
    ],
    "stream": false
}

Response · 200

{
    "id": "chatcmpl-\u2026",
    "object": "chat.completion",
    "choices": [
        {
            "index": 0,
            "message": {
                "role": "assistant",
                "content": "Customer wants a quote for year-end accounts."
            },
            "finish_reason": "stop"
        }
    ],
    "usage": {
        "prompt_tokens": 24,
        "completion_tokens": 11,
        "total_tokens": 35
    }
}

GET https://models.bytepilot.ai/v1/models List the models you can call ⌄

The catalogue available through the Model API, in the same shape OpenAI clients expect — so a model picker built against OpenAI populates itself. Use the `id` from here as the `model` field in a completion.

Required scopemodels

Response · 200

{
    "object": "list",
    "data": [
        {
            "id": "google/gemini-3.1-flash-lite",
            "object": "model"
        }
    ]
}

Predictable failure modes

Errors

Both APIs use the same status codes. The Model API returns them in the error shape familiar model clients already understand.

401

The key is missing, malformed or revoked.

403

The key is valid but lacks the scope this endpoint needs.

402

Out of credit, or this key hit its own daily/monthly spend cap. Top up, or raise the cap on the Developers page.

404

No such record — or it belongs to another account. We never distinguish the two.

429

Too many requests. Back off and retry.

For tools and coding assistants

Machine-readable

Every format is generated from the same source as this page, so a human, client generator and coding assistant receive the same contract.

JSON/openapi.jsonOpenAPI 3.1 for generators and Postman TXT/llms.txtShort index for coding assistants TXT+/llms-full.txtThe complete reference in one file

Using an AI coding assistant? Point it at https://bytepilot.ai/llms.txt and ask it to build the integration.

Build without waiting

Create a scoped key and make the first call.

Your free credit works across the platform and both APIs.

Create your account