Nexith Core is now serving on our own sovereign GPU cluster — cache hits save up to 90%, OpenAI-compatible.Learn more →
Nexith
Sign Up
Nexith Core · Now available

Frontier intelligence,
from our lab to your product.

Nexith Core is our general-purpose model — built for reasoning, writing, and code. Trained from the ground up and served through a clean, OpenAI-compatible API.

Try Nexith CoreRead the research
Text

Nexith Core

nexith-core

One text model for everything. Codes, runs long, remembers — and escalates hard problems on its own, so you never model-shop.

CODELONG-HORIZONMEMORYAUTO-ESCALATE
Input
Think
Output
→ Unified Reasoning →
Try Nexith Core

Nexith Image

Highest-fidelity image generation for when the result has to be right — crisp detail, faithful prompts, production-ready output.

HIGH FIDELITY1024²PROMPT-FAITHFUL
V0.00022 稳定nexith-image

Nexith Image Fast

Faster, lower-cost image generation for drafts, ideation, and volume — great results in a fraction of the time and spend.

FASTLOW-COSTDRAFTS & VOLUME
V0.00022 稳定nexith-image-fast
Drop into your existing stack
Python SDKTypeScriptLangChainLlamaIndexOpenAI SDKVercel AI SDKDeno
Nexith Core · Now available

Production-grade performance, priced for builders.

1M
Context window
0
Tokens / sec
−90%
On cache hits
SSE
Streaming on all endpoints
Why Nexith Core

Built for real work, not just demos.

Most models are impressive for a paragraph. Nexith Core is engineered to hold up across a full task — writing code, running long, remembering what matters, and answering instantly when it can.

01

Serious coding

Writes, reads, and refactors code across languages — and knows when a task is done. Trained to close the loop instead of spinning, so it ships working changes, not endless drafts.

MULTI-LANGUAGE · KNOWS WHEN TO STOP
02

Long-horizon tasks

Breaks a goal into steps, tracks its own budget, and drives to a clean finish. Stays coherent across long, multi-step work instead of losing the thread halfway through.

DECOMPOSE · BUDGET · CLOSURE
03

Persistent memory

Remembers across sessions — pin what matters by hand, or let it capture context automatically. Scoped per user and per app, so it picks up exactly where you left off.

CROSS-SESSION · SCOPED
04

Semantic cache

Recognizes when a request is close enough to one it has answered before and returns instantly — near-zero latency and dramatically lower cost, with cached input billed at a fraction of the usual rate.

INSTANT REPLAY · LOWER COST
Coding

Code that ships, not just compiles.

Nexith Core — it writes the change, runs the reasoning, and knows when to stop.

MULTI-LANGUAGEREAD · WRITE · REFACTORKNOWS WHEN DONEOPENAI-COMPATIBLE
API Playground
REQUEST
POST /v1/chat/completions Authorization: Bearer nx-•••••••• "model": "nexith-core" "messages": [{...}]
RESPONSE
200 OK · streaming "content": "Sure! Here's how..." "tokens": { prompt: 124, completion: 89 }
Long-horizon

Long tasks, start to finish.

Nexith Core stays coherent across dozens of steps — decomposing, budgeting, and closing out cleanly.

STEP DECOMPOSITIONSELF-BUDGETING1M CONTEXTCLEAN CLOSURE
Task run — 4 / 5 steps
On track
4
Steps done
62%
Budget used
18K
Context
Decompose goal into steps
Gather context & data
Draft solution
Review & refine
Close out cleanly
Memory + Cache

Remembers. And answers instantly.

Nexith Core carries context across sessions and replays close requests in milliseconds.

CROSS-SESSION MEMORYSEMANTIC CACHESCOPED PER USERLOWER COST
memory + cache
# Second call — same question, different session response = client.chat.completions.create( model="nexith-core", ... ) # Recalled from memory, served from cache cache: "hit" similarity: 0.97 served: "instant" cost: "−90%"
Simple, usage-based pricing

Pay only for what you generate.

Save up to 90% on cache hits, transparent per-token rates, and a free tier to start. No seats, no minimums.

View pricing
Pricing

Simple, transparent pricing

No hidden fees. Pay only for what you use.

Free
$0/month

Perfect for exploration

Get Started Free
  • 1M tokens free/month
  • Rate limited: 10 RPM
  • Community support
  • Nexith Core access
POPULAR
Pro
$20/month

For serious builders

Start Free Trial
  • 10M tokens/month
  • 300 RPM limit
  • Priority support
  • 1M context window
  • Analytics dashboard
Enterprise
Custom

For teams at scale

Contact Sales
  • Unlimited tokens
  • Custom rate limits
  • Dedicated support
  • SLA guarantee
  • Fine-tuning API
  • Private deployment

Input $4.00 · Cached input $0.40 · Output $20.00 per million tokens · No credit card required

Start building with Nexith Core

Join the developers shipping the next wave of AI products.

Get API Key — It's FreeRead the docs

No credit card required · Free tier available · Cancel anytime

Nexith — Frontier AI Models | Nexith