Inference.

Big models. Small bills.

$1 in free creditsFrom $0.10 per 1M input tokens

import osfrom openai import OpenAI client = OpenAI(    base_url="https://model.graphn.ai/v1",    api_key=os.environ["GRAPHN_API_KEY"],) reply = client.chat.completions.create(    model="deepseek-v4.1-flash",    messages=[{"role": "user", "content": "Hello"}],)print(reply.choices[0].message.content)

graphn.ai

graphn.ai · Inference

DeepSeek V4.1 Flash1M context

Kimi K2.5256K context

Mistral Large 3256K context

Qwen3 VL 235B256K context

Qwen3.8 27B256K context

+ Your model$ graphn model deploy

Managed inference · OpenAI-compatible API

Model launches

Oct 19, 2026DeepSeek V4.1 FlashLong-context reasoning, coding and agents, with image input Sep 28, 2026Kimi K2.5 (image)Frontier-scale agentic work with images Sep 28, 2026Qwen3 VL 235B Instruct (image)Demanding image understanding and visual agents Sep 28, 2026Mistral Large 3Long-context assistants and enterprise agents Aug 14, 2026Qwen3.8 27B (image+video)Coding, long-horizon agents and image/video in one dense model Apr 20, 2026Qwen3-VL Embeddings (4096 dim, multimodal)Puts text, images, screenshots and video in one vector space Apr 20, 2026Qwen3-VL Reranker (multimodal)Scores relevance across modalities, text against images and back Apr 20, 2026Gemma 4 E4B (text+image+video+audio)The catalogue's audio-in model - speech and images without a separate ASR step

The built-in catalog

Chat & vision · 10

Model	Context	Thinking	JSON	Vision	Audio	Input · 1M	Output · 1M
DeepSeek V4.1 Flash`deepseek-v4.1-flash`	1M				·	$0.30	$1.20
Kimi K2.5 (image)`kimi-k2.5`	256K	·	·		·	$0.60	$3.00
Mistral Large 3`mistral-large-3`	256K	·	·		·	$0.50	$1.50
Qwen3 VL 235B Instruct (image)`qwen3-vl-235b`	256K	·	·		·	$0.53	$2.66
Qwen3.8 27B (image+video)`qwen3.8-27b`	256K				·	$0.45	$3.20
gpt-oss 120B`gpt-oss-120b`	128K			·	·	$0.15	$0.60
Qwen3 80B Instruct`qwen3-80b`	128K	·		·	·	$0.20	$0.60
Qwen3.5 9B Vision (image+video)`qwen3-vl`	256K				·	$0.10	$0.15
Gemma 4 E4B (text+image+video+audio)`gemma-4-e4b`	64K		·			$0.15	$0.60
GLM-5.3`glm-5.3`	1M		·	·	·	$1.40	$4.40

Retrieval · 4

Model	Kind	Input · 1M
BGE Reranker v2`bge-reranker`	Text reranker	—
BGE-M3 Embeddings (1024 dim)`bge-m3`	Text embeddings	—
Qwen3-VL Embeddings (4096 dim, multimodal)`qwen3-vl-embedding`	Text and image embeddings	—
Qwen3-VL Reranker (multimodal)`qwen3-vl-reranker`	Text and image reranker	—

Every model has its own page with strengths, caveats, and a runnable request. The live list is in the app under Models, or via graphn model list-chat.

Swap the model. Keep the code.

Your weights. Our muscle.

Deploy weights from Hugging Face or S3, LoRA included, behind the same API.

Custom model imports →

graphn.ai

import_model.py

import graphn

Wassily Kandinsky, Circles in a Circle (1923)

Let the prompt pick the model

Send auto and each request goes to a model that can handle it, or build your own router. You pay for the model that answers.

graphn.ai

app.py

reply = client.chat.completions.create(
    model="auto",
    messages=messages,
)

See every request

Model, key, status, time to first token, tokens and cost for every call.

graphn.ai

Call it from anywhere

OpenAI's request shape, so the tooling you already have works unchanged.

Models documentation →

terminal

curl https://model.graphn.ai/v1/chat/completions \
  -H "Authorization: Bearer $GRAPHN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-oss-120b",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

app.ts

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://model.graphn.ai/v1",
  apiKey: process.env.GRAPHN_API_KEY,
});

const reply = await client.chat.completions.create({
  model: "deepseek-v4.1-flash",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(reply.choices[0].message.content);

quickstart.py

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://model.graphn.ai/v1",
    api_key=os.environ["GRAPHN_API_KEY"],
)

reply = client.chat.completions.create(
    model="qwen3-80b",
    messages=[{"role": "user", "content": "Hello"}],
)
print(reply.choices[0].message.content)

Paul Klee, Fish Magic (1925)