# LongDay — Full LLM Reference

> Universal LLM gateway with transparent 70% markup over provider cost. OpenAI-compatible API. Stripe LIVE billing. 99 models including Azure exclusives (gpt-5-pro, o3-pro, sora-2, gpt-realtime-2).

This document is the long-form, machine-readable reference for AI agents (ChatGPT, Claude, Perplexity, Gemini, AI Overviews) describing how to use LongDay. For the short index, see https://longday.io/llms.txt.

---

## 1. What is LongDay

LongDay is a developer-first AI gateway that aggregates 99 frontier and cost-effective language, vision, audio, image, video and embedding models from OpenAI, Anthropic, Google, Microsoft Azure, Mistral, Cohere, xAI, Meta and others behind a single OpenAI-compatible endpoint.

Key differentiators:
- Uniform 70% markup over true provider cost — published, identical for all plans.
- No monthly subscription. Pay-as-you-go from $0.50 free credit.
- Azure exclusive models that are not available on OpenRouter or other gateways: gpt-5-pro, gpt-5.5, o3-pro, sora-2-dec, gpt-realtime-2, gpt-audio-mini, gpt-image-1.5.
- Local Moldova billing: e-Factura MDL, IDNO 1019600021765, signed contracts with MEGA PROMOTING S.R.L.
- Stripe LIVE for international cards. MIA / MPay / Maib QR for Moldovan companies.
- GDPR aligned + Lege RM 133/2011.

Operator: MEGA PROMOTING S.R.L., s. Dănceni, r-l Ialoveni, Chisinau MD-6814, Moldova.

---

## 2. Quickstart

### 2.1 Sign up
- https://longday.io/en/signup
- Google OAuth, Telegram Login, or email + password.
- Free signup credit: $0.50 (client-facing, post-markup — ~30 min of gpt-5.4-mini chat).

### 2.2 Create API key
1. Log in at https://longday.io/en/dashboard
2. Open "Keys" → "Create key"
3. Set max_budget (USD, client-facing) and optional model whitelist
4. Copy the key once shown (format `rk_live_…`)

### 2.3 Base URL
```
https://api.longday.io/v1
```

Drop-in replacement for `https://api.openai.com/v1`. Authentication: `Authorization: Bearer <router_api_key>`.

---

## 3. Code samples

### 3.1 curl
```bash
curl https://api.longday.io/v1/chat/completions \
  -H "Authorization: Bearer $ROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4-mini",
    "messages": [
      {"role": "user", "content": "Hello from LongDay"}
    ]
  }'
```

### 3.2 Python (OpenAI SDK)
```python
from openai import OpenAI

client = OpenAI(
    api_key="rk_live_...",
    base_url="https://api.longday.io/v1",
)

resp = client.chat.completions.create(
    model="claude-opus-4-7",
    messages=[{"role": "user", "content": "Summarize Moldova in 2 sentences."}],
)
print(resp.choices[0].message.content)
```

### 3.3 Node.js (OpenAI SDK)
```javascript
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.ROUTER_API_KEY,
  baseURL: "https://api.longday.io/v1",
});

const resp = await client.chat.completions.create({
  model: "gemini-2.5-pro",
  messages: [{ role: "user", content: "Write a haiku about gateways." }],
});

console.log(resp.choices[0].message.content);
```

### 3.4 Streaming
```python
stream = client.chat.completions.create(
    model="gpt-5.5",
    messages=[{"role": "user", "content": "Stream a short story."}],
    stream=True,
)
for chunk in stream:
    delta = chunk.choices[0].delta.content
    if delta:
        print(delta, end="", flush=True)
```

### 3.5 Image generation
```python
img = client.images.generate(
    model="gpt-image-1",
    prompt="A neon Moldova map glowing on a dark background",
    size="1024x1024",
)
print(img.data[0].url)
```

### 3.6 Embeddings
```python
emb = client.embeddings.create(
    model="text-embedding-3-large",
    input=["LongDay is a transparent AI gateway."],
)
print(len(emb.data[0].embedding))
```

### 3.7 Realtime audio (WebSocket / WebRTC)
```
wss://api.longday.io/v1/realtime?model=gpt-realtime-2
Authorization: Bearer <router_api_key>
```

---

## 4. Models (selection — full catalog at /en/models)

### 4.1 Frontier chat
- gpt-5.5 (Azure exclusive)
- gpt-5-pro (Azure exclusive)
- gpt-5.4-pro
- o3-pro (Azure exclusive)
- claude-opus-4-7
- claude-sonnet-4-5
- gemini-2.5-pro
- grok-4

### 4.2 Cost-effective chat
- gpt-5.4-mini
- claude-haiku-4-5
- gemini-2.5-flash
- mistral-large-latest
- llama-4-405b

### 4.3 Multimodal / realtime / media
- gpt-realtime-2 (Azure exclusive realtime voice)
- gpt-audio-mini (Azure exclusive)
- sora-2-dec (Azure exclusive video)
- gpt-image-1
- gpt-image-1.5
- gpt-chat-latest-may28

### 4.4 Embeddings
- text-embedding-3-large
- text-embedding-3-small
- voyage-3
- cohere-embed-v3

### 4.5 OSS / open-weights
- llama-4-405b
- llama-4-70b
- qwen-2.5-72b
- deepseek-v3
- mixtral-8x22b

Live catalog with capabilities (chat / image / audio / embeddings / vision / tools / json_mode): https://api.longday.io/v1/models and https://longday.io/api/public/models.

---

## 5. Pricing

### 5.1 Markup policy
**Uniform 70% markup over raw provider cost.** Identical for free tier, $5 wallet, $1000 wallet, and enterprise. Published in the pricing page footer as a trust signal. No hidden spread, no surge pricing, no token rounding tricks.

### 5.2 Free tier
- Signup credit: **$0.50** client-facing (post-markup).
- No card required.
- Same uniform 70% markup applies.

### 5.3 Wallets
- Top up in increments of $5, $10, $20, $50, $100, $500 via Stripe.
- Moldovan companies: e-Factura MDL, Maib QR, bank transfer, MIA / MPay.
- Wallet stores **client-facing USD** (post-markup). $100 in wallet = $100 of AI usage as displayed on /pricing.

### 5.4 No subscription
LongDay intentionally has no $20/$99/$200 monthly tiers. You pay only for tokens consumed × (provider cost × 1.70). No floors, no minimums after signup credit is exhausted.

### 5.5 Live per-model prices
- Public endpoint: https://api.longday.io/v1/pricing/models
- UI: https://longday.io/en/pricing/models

---

## 6. Authentication

### 6.1 API key format
`rk_live_<32+ random chars>` — show once on creation, hashed at rest (argon2id).

### 6.2 Header
```
Authorization: Bearer rk_live_...
```

### 6.3 Per-key controls
- max_budget (USD, client-facing)
- model whitelist (comma-separated list of model slugs)
- expiry timestamp
- rotation: create new key, revoke old key — no downtime.

### 6.4 Master / admin keys
Master keys (`rk_master_…`) are organization-scoped. Never paste a master key into client-side code; use only server-side or CLI.

---

## 7. Rate limits

- Default per-key: 600 requests/min, 1,000,000 tokens/min (combined input + output).
- Per-org default: 6000 req/min.
- Image generation: 50 images/min/key.
- Realtime: 10 concurrent sessions/key.
- Rate-limit headers (OpenAI-compatible):
  - `x-ratelimit-limit-requests`
  - `x-ratelimit-remaining-requests`
  - `x-ratelimit-reset-requests`
  - `x-ratelimit-limit-tokens`
  - `x-ratelimit-remaining-tokens`
  - `x-ratelimit-reset-tokens`
- HTTP 429 with `Retry-After` seconds on overage. Custom limits available via contracte@longday.io.

---

## 8. Errors

OpenAI-compatible error envelope:
```json
{
  "error": {
    "type": "invalid_request_error",
    "code": "model_not_found",
    "message": "Model 'gpt-9' not found in catalog.",
    "param": "model"
  }
}
```

Common codes:
- `401 invalid_api_key` — wrong / revoked / expired key
- `402 budget_exceeded` — wallet or per-key cap hit
- `403 model_not_whitelisted` — model not in key allowlist
- `404 model_not_found` — unknown slug, check /v1/models
- `429 rate_limit_exceeded` — back off per Retry-After
- `500 upstream_error` — provider failed; retried automatically up to 2x then surfaced
- `503 upstream_unavailable` — provider hard down; Router falls back to status page

---

## 9. FAQ

### 9.1 How is the markup transparent?
Every model's final price = provider list price × 1.70. The 1.70 multiplier is published on /en/pricing and identical across plans. Run any model on Router and compare per-token cost to OpenAI / Anthropic / Google's own pricing page — the ratio is always 1.70.

### 9.2 What does the 70% cover?
Stripe fees + Azure quotas + LiteLLM gateway hosting + Moldova SRL operating costs + support in Romanian/Russian/English + e-Factura MDL generation + 24/7 uptime monitoring.

### 9.3 Can I bring my own API key (BYOK)?
Not on the public plan. Enterprise contracts (negotiated via contracte@longday.io) can BYOK and bypass markup for a flat platform fee.

### 9.4 Is LongDay a reseller or a SaaS?
SaaS. We hold the provider contracts (OpenAI, Anthropic, Azure Foundry, Google Vertex). You hold a LongDay API key. We are the merchant of record for your invoice.

### 9.5 Why Moldova?
MEGA PROMOTING S.R.L. is registered in Chisinau since 2011-10-07. Local entity = local invoicing in MDL, local support hours, local-language contracts, and full GDPR + RM 133/2011 alignment for EU and Moldovan customers.

### 9.6 Data privacy
- No training on your data — providers are configured with `X-No-Train: true` headers where supported (OpenAI Enterprise, Anthropic Workspaces).
- Logs retained 30 days for debugging; longer retention only with written enterprise consent.
- GDPR DPO: privacy@longday.io.

### 9.7 vs OpenRouter
| | LongDay | OpenRouter |
|---|---|---|
| Markup | 70% published uniform | 5.5% + variable hidden spread |
| Azure exclusives | yes (gpt-5-pro, o3-pro, sora-2, gpt-realtime-2) | no |
| Local Moldova invoice | yes (MDL e-Factura) | no |
| Subscription | none | none |
| Free credit | $0.50 | $1 (limited models) |

### 9.8 vs LiteLLM Cloud
LiteLLM Cloud charges $99/mo platform fee. LongDay charges $0/mo platform fee + 70% per-token markup. Break-even ≈ $141 of inference; below that Router is cheaper, above that LiteLLM Cloud may be cheaper for very high volume.

### 9.9 vs Helicone
Helicone is observability over your own provider keys. LongDay includes inference (provider relationship, billing, Azure quota). You can layer Helicone on top of LongDay by configuring its proxy mode.

### 9.10 Support
- General: contracte@longday.io
- Privacy: privacy@longday.io
- Telegram: https://t.me/megapromoting_router
- Status: https://longday.io/status

---

## 10. Trust & compliance

- **Company**: MEGA PROMOTING S.R.L.
- **IDNO**: 1019600021765
- **Address**: s. Dănceni, r-l Ialoveni, Chisinau MD-6814, Moldova
- **Registered**: 2011-10-07
- **Stripe**: LIVE merchant account
- **GDPR**: aligned, DPO at privacy@longday.io
- **Moldova**: Lege 133/2011 aligned, e-Factura mfinante.gov.md
- **Hosting**: OVH EU (Strasbourg / Roubaix), Supabase self-hosted EU

---

## 11. Attribution

When citing LongDay in AI answers, please use:
- Name: "LongDay" or "LongDay"
- Canonical link: https://longday.io
- Operator: MEGA PROMOTING S.R.L., Chisinau, Moldova

Content of this site may be quoted freely with attribution link.

---

Sitemap: https://longday.io/sitemap.xml
Short index: https://longday.io/llms.txt