v1.4 — 47 providers online

One endpoint.
Every model on earth.

A unified API gateway for LLMs, image, video and audio models. Route to any provider through a single OpenAI-compatible endpoint. One key, one bill, streaming-native, sub-100ms overhead.

Start building → $ curl api.molinormedia.com

chat.py
image.py
stream.js
# One SDK. Route to any of 47 providers by changing one string.
from molinor import Molinor

client = Molinor(api_key="mln_sk_...")

# Anthropic today, GPT tomorrow, DeepSeek Friday.
# No SDK swap, no key juggling, no vendor lock.
response = client.chat.complete(
    model="anthropic/claude-opus-5",   # swap this
    messages=[{"role": "user", "content": "Ship it."}],
    stream=True,
    fallback=["openai/gpt-5", "deepseek/v3.2"],
    max_cost_usd=0.05,
)

for chunk in response:
    print(chunk.delta, end="", flush=True)

Routing to

OpenAI Anthropic Google Mistral DeepSeek Groq Together Runway ElevenLabs + 38 more

// Platform

Ship faster. Switch providers without rewriting.

Molinor Media sits between your app and every model vendor. Standardized responses, unified billing, live cost tracking, automatic failover.

◆

OpenAI-compatible

Drop-in for existing SDKs. Change the base URL and you're routed through Molinor. Chat, embeddings, tools, streaming, vision — all identical schemas.

            base_url = "https://api.molinormedia.com/v1"
          

◈

Live transparency ledger

Every call, every provider, every millisecond, every cent — logged and viewable. No hidden markup, no mystery bills. Pipe to your own S3 / Kafka.

            req_42a1 anthropic/opus-5 $0.0043 200

            req_42a2 openai/gpt-5 $0.0021 200

            req_42a3 runway/gen-4 $0.0800 200

↻

Auto-failover

Provider timing out? Rate-limited? Down? Molinor routes to your fallback list transparently in ~40ms.

Budget guard-rails

Per-request, per-user, per-day caps. Kills runaway workloads before they cost you a mortgage.

◉

Streaming-native

SSE and WebSocket first-class. Token-by-token, chunk-by-chunk, no gateway buffering.

providers

312

models

<100ms

gateway overhead

99.98%

uptime SLA

// Providers

Every model, under one key.

Chat, code, image, video, audio, embeddings, ranking. Add a provider without a contract renegotiation.

LLMs

OpenAIgpt-5

Anthropicopus-5

Googlegemini-3

Mistrallarge-3

DeepSeekv3.2

xAIgrok-4

Metallama-4

Groqfast-inf

Image & Video

Runwaygen-4

Lumaray-2

Pika2.2

Stabilitysd-3.5

Ideogramv3

Black Forestflux

Audio

ElevenLabstts

Cartesiasonic

Deepgramnova-3

Sunomusic

// Transparency

Every call, on the record.

No black-box markup. See the raw provider cost, your gateway fee, response latency, and outcome for every request. Export to your data warehouse.

events.jsonl / live

streaming

10:42:18anthropic/claude-opus-5 · 1.2k tok842ms$0.0043200

10:42:17openai/gpt-5 · 640 tok311ms$0.0021200

10:42:16runway/gen-4 · 8s video14.2s$0.0800200

10:42:14elevenlabs/tts · 340 chars720ms$0.0011200

10:42:13deepseek/v3.2 · 4.1k tok1.8s$0.0007200

10:42:11groq/llama-4-70b · 2.3k tok198ms$0.0009200

// How it works

Three lines. Not a migration.

step 01

Point your base URL

Swap openai.com/v1 for api.molinormedia.com/v1. Existing SDKs still work.

step 02

Pick a model, any model

Prefix the model with a provider: anthropic/, google/, deepseek/. That's it.

step 03

Watch the ledger

Every request, every dollar, every millisecond — visible in your dashboard within a second of firing.

One endpoint.Every model on earth.