openai · Language
GPT-OSS 120B
openai/gpt-oss-120b
MoE · 117B params · 5.1B active
- Context
- 128K
- LoRA rank
- 32
- Fine-tune
- $0.68
- Serve
- $0.45
Nucleus builds and runsyour enterprise LLM
We fine-tune models on your data and deploy them as private API endpoints, with dashboards, automation, scheduling and monitoring built in.
Your secure LLM, at ChatGPT-level intelligence, with full data privacy.
Trusted by leading institutions
Product
The platform for models you own
Fine-tune open models on your own data, serve them privately, and keep the weights. Three meters run the bill: the tokens you train, the tokens you serve, and the storage you keep.
33 open models, five modalitiesFine-tuning that goes
Fine-tuning that goes
as deep as you need
Language, vision, image, video and audio bases. Start with a managed job and go all the way down to the optimiser step when the task asks for it.
- Managed jobs. Upload a JSONL file of examples, pick a base model, start a run. Scheduling, GPUs and the production handoff are ours.
- Full control of the recipe. Your base model, your data, your method, with LoRA ranks up to 64 and evaluations billed on the same meters.
- The training loop itself. Forward-backward passes and optimiser steps are first-class API calls, so RL, DPO and distillation are yours to write.
dataset.jsonlLoRA r=32step 1,240 / 2,000From the first call to production trafficPrivate endpoints your
Private endpoints your
SDK already speaks
Every checkpoint deploys to a private endpoint the moment training finishes. Point your existing OpenAI or Anthropic client at a new base URL and change nothing else.
- Serverless. A fine-tuned model is served at its base model's rate. Idle time costs nothing.
- Dedicated. Reserve a GPU by the second, from L40S to B200, and run unlimited tokens on it. Cold starts are not charged.
- Reserved. One to six month commitments run 15 to 25% below on-demand, with capacity held for you.
api.nucleus-ai.ioOpenAI compatibleautoscale to zeroWhat you keep when you leaveThe model is yours
The model is yours
to take anywhere
Training data stays in your account and trains nobody else's models. Export any checkpoint through the API and run it on hardware you control.
- Weight export. Any checkpoint, any time, through the API. The model runs wherever you put it.
- Data residency. Singapore by default, your own cloud account, or hardware inside your building.
- On-premise. A platform licence from $50,000 a year, with your GPU and cloud infrastructure billed separately.
checkpoint exportregion ap-southeast-1runs air-gappedModels
Open models, ready to fine-tune
Language, vision, image, video and audio bases. Train any of them on your own data, serve them behind an OpenAI-compatible endpoint, and export the weights whenever you want them.
Qwen · Vision
Qwen3-VL 235B Instruct
Qwen/Qwen3-VL-235B-A22B-Instruct
MoE · 235B params · 22B active
- Context
- 128K
- LoRA rank
- 32
- Fine-tune
- $2.40
- Serve
- $1.60
black-forest-labs · Image generation
FLUX.2 dev
black-forest-labs/FLUX.2-dev
Flow transformer · 32B params
- Context
- Up to 4MP
- LoRA rank
- 32
- Fine-tune
- $1.40
- Serve
- $8.00
Lightricks · Video generation
LTX-2
Lightricks/LTX-2
DiT · native audio + video
- Context
- Up to 4K · 10s
- LoRA rank
- 32
- Fine-tune
- $1.60
- Serve
- $5.50
deepseek-ai · Language
DeepSeek V4 Pro
deepseek-ai/DeepSeek-V4-Pro
MoE · 1.6T params · 49B active
- Context
- 1M
- LoRA rank
- 16
- Fine-tune
- $5.60
- Serve
- $3.50
google · Vision
Gemma 4 31B
google/gemma-4-31B
Dense · 31B params
- Context
- 256K
- LoRA rank
- 32
- Fine-tune
- $1.45
- Serve
- $0.95
Products
Own the whole journey
From dataset to deployed model: integrations that make your data trainable, managed GPUs, private endpoints and the APIs that tie them together.
Integrations
Connect your databases, APIs and document stores and turn them into training-ready datasets for fine-tuning.
- Databases, APIs and document stores
- Training-ready datasets
- Custom integration support
AI Cloud
Train on managed GPUs and serve your models without deploying or scaling servers.
Custom AI Solutions
Fine-tuned models built on your enterprise data, evaluated and served as private endpoints.
- OpenAI + Anthropic compatible APIs
AI Chat Agents
Assistants for your teams, backed by your own fine-tuned models rather than a shared public model.
APIs for custom solutions
Integrate your models into existing products with clean REST APIs, documented end to end.
Use cases
Built with Nucleus
Explore demos our team has built with Nucleus across different industries.
Ad Performance Tracking
"Our Q4 campaigns across Google AdWords and Facebook are showing inconsistent performance. Can you analyze our Matomo user data and Stripe revenue to optimize our advertising spend and improve attribution?"
Shipment Delay Management
"Check if any shipments have been delayed more than 2 days past ETA. Update chat channel and email relevant managers."
Vendor Performance Analytics
"I need a comprehensive vendor performance analysis for our Q4 contract renewals. Focus on our top 10 vendors and include cost efficiency, service quality, and response times."
Intelligent Fleet Optimisation
"Our delivery trucks are consistently running late and fuel costs are through the roof. Analyze our GPS tracking data, optimise routes and automatically update customers when delays happen."
Agriculture Yield Analysis
"Why is rice yield forecast low in Chiang Mai Block 4B this season? I need to understand what went wrong and prevent it from happening in other blocks."
Critical Incident Response
"Are there any critical incidents that have been unresolved for more than 4 hours. Escalate to senior engineers and update our incident response channels."
Dynamic Feeding Schedules
"I need to optimize our tiger feeding schedules based on their health data, activity patterns, and environmental conditions. Can you analyze the data and create personalized feeding plans?"
Predictive Maintenance
"Our production line 3 has been experiencing unexpected downtime. Can you analyze our equipment sensor data, predict potential failures, and automatically coordinate maintenance schedules to prevent costly production stops?"
Quickstart
From dataset to a model you own
A handful of API calls take a JSONL file to a fine-tuned model served behind an OpenAI-compatible endpoint. Prefer to drive the training loop yourself? Forward-backward passes and optimiser steps are first-class API calls too.
nucleus quickstart
01_upload_dataset.sh
Bring a JSONL file of examples. That's the only prerequisite.
System01 / 05
Teams
Private models for every team
Fine-tune, evaluate, serve, and export models your organisation owns, with a workflow for every team that touches them.
ML Engineers
Fine-tune and deploy custom models on your enterprise data with full control over training pipelines and inference endpoints.
Research Teams
Drive the training loop call by call. RL, DPO, distillation, and custom losses, with the GPUs handled for you.
Product Teams
Ship features on models that speak your domain. Keep your existing OpenAI or Anthropic client code and swap the base URL.
Data Teams
Turn proprietary datasets into evaluated, versioned models with managed fine-tuning jobs. Upload a JSONL file, get back a private endpoint.
Platform Teams
Serve models behind OpenAI- and Anthropic-compatible endpoints, with usage tracking and rate limits built in.
Enterprise Leaders
Own your AI strategy. Training data stays in your account, weights are exportable, and models can run wherever policy requires.
Scale
Built to scale. Built to be owned.
Nucleus handles training and inference at any scale, from the first fine-tuning job to full production traffic, so your team ships models instead of managing GPUs.
- Your model weights
100% yours
- Compatible APIs
OpenAI + Anthropic
Tokens served
10.9M
+13.4%