One gateway for 100+ LLMs.

Use any model through Onyx and track usage and spend in one place.

Book a Demo

Read the docs

wire protocols

endpoint, one bill

provider keys on clients

Claude Code

Codex CLI

Cursor

Your app

Onyx Gateway

/api/gateway/v1

Metered spend

$4.2130

OpenAI

Anthropic

Gemini

Azure

Bedrock

DeepSeek

100+ models.

Zero new API keys.

Claude Fable 5

Claude Opus 4.8

Claude Sonnet 5

GPT-5.6 Sol

GPT-5.6 Terra

GPT-5.6 Luna

GPT-5.5

Gemini 3.5 Flash

Gemma 4 31B

Gemma 4 12B

Muse Spark 1.1

Grok 4.5

Mistral Small 4

Mistral Medium 3.5

Command A+

DeepSeek-V4-Pro

DeepSeek-V4-Flash

Qwen3.6-35B-A3B

Qwen3.7-Max

Qwen3.6-27B

Kimi K3

GLM-5.2

MiniMax M3

MiniMax M2.7

Step-3.7-Flash

MiMo-V2.5-Pro

Hunyuan Hy3

Nemotron 3 Ultra

Nemotron 3 Super

Nemotron 3 Nano

ERNIE 5.1

Ling-2.6-1T

Nova 2 Pro

Kimi K2.6

GLM-4.7

Qwen 3.5

DeepSeek R1

Gemma 3 27B

Llama 4 Maverick

Claude Fable 5

Claude Opus 4.8

Claude Sonnet 5

GPT-5.6 Sol

GPT-5.6 Terra

GPT-5.6 Luna

GPT-5.5

Gemini 3.5 Flash

Gemma 4 31B

Gemma 4 12B

Muse Spark 1.1

Grok 4.5

Mistral Small 4

Mistral Medium 3.5

Command A+

DeepSeek-V4-Pro

DeepSeek-V4-Flash

Qwen3.6-35B-A3B

Qwen3.7-Max

Qwen3.6-27B

Kimi K3

GLM-5.2

MiniMax M3

MiniMax M2.7

Step-3.7-Flash

MiMo-V2.5-Pro

Hunyuan Hy3

Nemotron 3 Ultra

Nemotron 3 Super

Nemotron 3 Nano

ERNIE 5.1

Ling-2.6-1T

Nova 2 Pro

Kimi K2.6

GLM-4.7

Qwen 3.5

DeepSeek R1

Gemma 3 27B

Llama 4 Maverick

GPT-oss 120B

Nemotron Ultra 253B

DeepSeek V3.2

Nemotron Super 49B

Step-3.5-Flash

Step3

MiMo-V2-Flash

Nemotron Nano 30B

Mistral Large 3

Llama 3.3 70B

DS-R1-Distill-Llama-70B

Llama 4 Scout

Hunyuan 2.0

Qwen3-Coder-Next

Mistral Small 3.1

GPT-oss 20B

Phi-4

Phi-4-mini

Gemma 3 12B

DS-R1-Distill-Qwen-32B

DS-R1-Distill-Qwen-14B

Gemini 3.1 Pro

Grok 3

Devstral-2-123B

Qwen3.5-35B-A3B

Qwen3.5-27B

Qwen3.5-122B-A10B

Qwen3.5-9B

Qwen3.5-4B

Ministral 14B

Ministral 8B

Ministral 3B

GLM-Z1-32B

GLM-Z1-9B

DeepSeek-R1-0528

DeepSeek-R1-0528-Qwen3-8B

Kimi-Linear-48B-A3B

Qwen3-Coder-480B-A35B

GPT-oss 120B

Nemotron Ultra 253B

DeepSeek V3.2

Nemotron Super 49B

Step-3.5-Flash

Step3

MiMo-V2-Flash

Nemotron Nano 30B

Mistral Large 3

Llama 3.3 70B

DS-R1-Distill-Llama-70B

Llama 4 Scout

Hunyuan 2.0

Qwen3-Coder-Next

Mistral Small 3.1

GPT-oss 20B

Phi-4

Phi-4-mini

Gemma 3 12B

DS-R1-Distill-Qwen-32B

DS-R1-Distill-Qwen-14B

Gemini 3.1 Pro

Grok 3

Devstral-2-123B

Qwen3.5-35B-A3B

Qwen3.5-27B

Qwen3.5-122B-A10B

Qwen3.5-9B

Qwen3.5-4B

Ministral 14B

Ministral 8B

Ministral 3B

GLM-Z1-32B

GLM-Z1-9B

DeepSeek-R1-0528

DeepSeek-R1-0528-Qwen3-8B

Kimi-Linear-48B-A3B

Qwen3-Coder-480B-A35B

Everything a gateway should do

Three protocols. One base URL.

POST /gateway/v1/chat/completionsOpenAI Chat

POST /gateway/v1/responsesOpenAI Responses

POST /gateway/v1/messagesAnthropic Messages

Every provider you run

OpenAIAnthropicAzure OpenAIAmazon BedrockGoogle Vertex AIOpenRouterOllamavLLMLM StudioAny OpenAI-compatible endpoint

Budgets checked before the call

Token or dollar ceilings per workspace, group, or user.

Provider keys stay server-side

Clients hold a scoped Onyx token. Requests carry their permissions.

Usage and spend, in one place

Chat, agents, and gateway calls in one ledger. Search by user, drill into a day, export CSV or PDF.

Per user

daily detail

Per model

negotiated rates

Per flow

chat, agents, gateway

CSV + PDF

exports

Workspace usage

Last 30 days

Total spend

$4,218.40

Gateway share

61%

Active users

184

m.chen

$612.18

a.okafor

$498.05

s.patel

$377.90

j.laurent

$240.33

r.kim

$188.76

Gateway

Chat

Agents

Set up in minutes

Gateway docs

Book a Demo

Add models in Onyx

Access rules carry over.

Create a gateway token

Scoped. Revoke any time.

Point your tool at it

Or paste the agent prompt.

1Set up the Onyx LLM Gateway for this project.
2
31. Ask me for my Onyx gateway base URL (https://<my-onyx-domain>/api/gateway/v1)
4   and a gateway-scoped personal access token (onyx_pat_...). Never echo the token.
52. Call GET <base_url>/models with "Authorization: Bearer <token>" and list the
6   model IDs. They look like <provider-id>/<model-name>, for example 18/claude-sonnet-5.
73. Configure the tools I use to route through the gateway:
8   - Claude Code: ANTHROPIC_BASE_URL, ANTHROPIC_AUTH_TOKEN, ANTHROPIC_MODEL
9   - Codex CLI: OPENAI_BASE_URL, OPENAI_API_KEY
10   - OpenAI or Anthropic SDKs in this repo: base_url and api_key in the client config
11   Put secrets in the environment or a gitignored .env, never in committed files.
124. Send one short test request, confirm streaming works, and tell me which model answered.

Get Started

Today

Start your 14-day free trial of Onyx (no credit card required)

Try Onyx Cloud

Secure

Enterprise-grade security and compliance. Flexible and secure deployment options.

Open-Source

Deep customizability optimized for your use case, supported by a large Open-Source community.

Stars

20k

Docs

Trusted by

top teams

Case study