The independent AI gateway

One connection.
Every model.

Nexa gives your coding tools and applications one reliable API for the models you want to use. Build faster, change less, and keep the choice yours.

OpenAI-compatible Responses, Chat Completions & Anthropic Messages API
nexa / request live
01POSTapi.nexa.dev/v1/responses
02Authorization: Bearer nx_live_••••••••
03model: claude-sonnet-4
04input: “Ship something great.”
200 response 842ms
{ output :“Ready when you are.” }
Claudeconnected
GPTconnected
DeepSeekconnected
More room to build

A little credit.
A lot of possibility.

One token balance. Every model. More tokens as you go. Compare the published multipliers, then choose the route that fits your work.

Top-up amountUSD
$
$1$500 max
You receive
465.47Mbalance tokens
+4.73% volume bonus
Volume bonus
+4.73%
Price per 1M balance tokens
$0.0215

0% at $1. Increases evenly to +60% at $100, then more slowly to +80% at $150.

Top up in workspace

Final credit is confirmed securely after payment.

Every model. One balance.

Estimated price per 1M model tokens.
13 models · Live catalog
Model / multiplierInput / outputCached inputContext
GPT-6 Astra×8$0.17191.05M$0.1719
Claude Opus 5×5$0.10741M$0.1074
Claude Opus 4.8×4$0.08591M$0.0859
Claude Opus 4.7×4$0.08591M$0.0859
Claude Opus 4.6×4$0.08591M$0.0859
Claude Sonnet 5×2.5$0.05371M$0.0537
Claude Sonnet 4.6×2$0.04301M$0.0430
GPT-5.6 Sol×4$0.08591.05M$0.0859
GPT-5.5×4$0.08591.05M$0.0859
GPT-5.6 Terra×3$0.06451.05M$0.0645
GPT-5.6 Luna×1.7$0.03651.05M$0.0365
DeepSeek V4 Pro×2$0.04301M$0.0430
DeepSeek V4 Flash×0.8$0.01721M$0.0172

Your selected bonus lowers every model's effective price.

Base rate $0.0215 / 1M balance tokens · Loading live USD/EUR exchange rate…
CodexClaude CodeClaudeYour app
01 / The Nexa way

Less plumbing.
More making.

One thoughtful layer between your product and the models behind it. Simple enough to start today, flexible enough to grow with your work.

01

One place to connect

Set one base URL and one key. Send familiar requests through a single, clean gateway.

See the API
02

Your choice, always

Move between GPT, Claude and DeepSeek model families without rewriting your application.

View models
03

Clarity by default

See usage, balance and model multipliers in one calm workspace, before costs become surprises.

See pricing
02 / One continuous path

From thought
to response.

Your application keeps its familiar shape. Nexa handles the route, the model connection, and the return.

01Your appRequest
ROUTE
NEXA
RETURN
03Your modelResponse
03 / Familiar by design

Keep your stack.
Change the connection.

Use Nexa with the clients and workflows you already know. Add your key, point to our base URL, and choose from every model in the catalog.

01curl https://api.nexabase.xyz/v1/chat/completions \
02-H "Content-Type: application/json" \
03-H "Authorization: Bearer $NEXA_API_KEY" \
04-d '{
05"model": "gpt-5.6-terra",
06"messages": [
07{ "role": "user", "content": "Make something worth building." }
08],
09"temperature": 0.7,
10"stream": true
11}'
api.nexabase.xyz · ready
POST/v1/chat/completions
04 / Clear by design

Pricing that
stays readable.

Start with a simple balance. See exactly how each model uses it, then choose the route that makes sense for your work.

Base credit cost
$0.0225
per 1M balance tokens

No monthly commitment. Volume and launch credits apply automatically.

Explore model pricing
Claude Sonnet 4× 0.8
GPT-5.6× 1.0
DeepSeek× 0.6
Model multipliers determine how many balance tokens each request uses.
05 / A few details

Good questions.
Clear answers.

Read the full docs

Nexa is an independent AI gateway for developers. Use one compatible API to connect the models and coding tools you already work with, then switch models without rebuilding your integration.

Yes. Nexa follows familiar OpenAI-compatible request formats. Set your base URL, add your Nexa key, and choose a supported model from the catalog.

Nexa is designed for Codex, Claude Code, DeepSeek, and your own applications. Endpoint support depends on the selected model and protocol.

Add credit to your workspace and pay for the balance tokens your requests use. Each model has a clear multiplier so you can compare cost before you send.

Ready when you are

Make the connection.

A calmer way to bring more intelligence into the things you build.

Create your account