Novita AI
AI and agent cloud: 200+ serverless model APIs, dedicated endpoints, GPU instances and an agent sandbox
Novita AI serves 200+ models as serverless APIs covering LLMs, image, video, audio and vision, behind an OpenAI-compatible API. Dedicated endpoints give a model its own capacity. Batch inference runs at an introductory 50% discount on supported models.
The same account rents GPU instances by the hour, from $0.33/hour for an RTX 4090 (spot $0.17). It also runs Agent Sandbox: isolated runtimes where coding agents run code and use tools, billed per second for vCPU and memory.
Pricing: Usage-based
Data and privacy
Checked 17 November 2026- DPA
- Not public
- Sub-processors
- View the list โ
- US parent company
- Yes
Read from Novita AI's own legal pages. "Not public" means it is not on their site; some providers share it on request.
Choosing an inference API right now?
One click, anonymous. Helps us know our visitors.
Thanks, that helps.
Choosing an inference API right now?
One click, anonymous. Helps us know our visitors.
Thanks, that helps.
Novita AI prices by model
Per 1M tokens, read off Novita AI's own pricing page on the date shown.
| Model | Input / 1M | Cached / 1M | Output / 1M | Checked | Notes |
|---|---|---|---|---|---|
| DeepSeek V4 Pro | $1.32 | $0.044 | $3.96 | 14 Nov 2026 | Listed as DeepSeek V4 Pro 0813 |
| DeepSeek V4.1 Flash | $0.30 | $0.006 | $1.20 | 14 Nov 2026 | |
| DeepSeek V4.1 Flash 1017 | $0.30 | $0.006 | $1.20 | 17 Nov 2026 | |
| GLM 5.3 | $1.40 | $0.26 | $4.40 | 14 Nov 2026 | |
| GLM 5.3 Flash | $0.15 | $0.03 | $0.50 | 14 Nov 2026 | |
| Gemma 4 26B A4B | $0.13 | โ | $0.40 | 16 Nov 2026 | |
| Gemma 4 31B | $0.14 | โ | $0.40 | 16 Nov 2026 | |
| Kimi K3.1 Pro | $2.00 | $0.20 | $10.00 | 17 Nov 2026 | Cache write $2 (5 min) / $4 (1 h). |
| Llama 3.3 70B Instruct | $0.135 | โ | $0.40 | 14 Nov 2026 | |
| Qwen3.8 2.4T A95B | $2.00 | $0.25 | $6.00 | 14 Nov 2026 | |
| Qwen3.8 27B | $0.42 | $0.085 | $3.00 | 14 Nov 2026 | |
| gpt-oss-120b | $0.05 | โ | $0.25 | 14 Nov 2026 |
Resources
What is Novita AI?
Novita AI is a cloud for developers and AI agents. One account covers serverless model APIs, dedicated endpoints, GPU instances and Agent Sandbox, an isolated runtime where agents run code.
Model APIs
The API serves 200+ models across LLMs, image, video, audio and vision. It is OpenAI-compatible and billed per token or per generated output. The catalogue includes recent open-weight releases such as DeepSeek V4.1 Flash 1017 and Kimi K3.1 Pro; Infrabase tracks their verified prices across providers on the DeepSeek V4.1 Flash 1017 and Kimi K3.1 Pro pages.
Dedicated endpoints reserve capacity for one model, for steady production traffic. Batch inference runs at an introductory 50% discount on supported models.
Agent Sandbox
Agent Sandbox gives an agent an isolated environment to run code, use tools and work with files. Novita names coding agents, browser automation and data processing as the uses, with startup around 200 ms. Billing is per second: $0.0000098 per vCPU-second and $0.0000032 per GiB of memory per second. The first 60 GB of storage are included (November 2026).
GPU instances
GPUs rent by the hour on demand or as cheaper spot capacity. An RTX 4090 costs $0.33/hour on demand or $0.17 on spot (November 2026). Current rates for every GPU are on novita.ai/pricing.
Who uses it?
Teams building multimodal applications, integrating open models or running agents that need a safe place to execute code, without operating the infrastructure themselves.
Frequently Asked Questions
What is Novita AI?
Novita AI is a cloud platform for developers and AI agents. One account covers 200+ serverless model APIs, dedicated endpoints, GPU instances and Agent Sandbox, an isolated runtime for agents.
Which models does the Novita AI API cover?
LLMs, image, video, audio and vision models, more than 200 in total, through an OpenAI-compatible API. Dedicated endpoints reserve capacity for a single model when production traffic needs consistent performance.
What is Novita's Agent Sandbox?
An isolated environment where AI agents run code, use tools and work with files. Novita names coding agents, browser automation and data processing as typical uses. It is billed per second for vCPU and memory.
How much does Novita AI cost?
Pricing is usage-based. As of November 2026, DeepSeek V4.1 Flash 1017 costs $0.30 input and $1.20 output per 1M tokens; Kimi K3.1 Pro costs $2.00 and $10.00. Agent Sandbox costs $0.0000098 per vCPU-second plus $0.0000032 per GiB-second. GPU instances start at $0.33 an hour for an RTX 4090 on demand.
Who is Novita AI for?
Developers and teams running production AI workloads, especially multimodal applications, open-model integrations and AI agents that need isolated execution.
Novita AI Alternatives
Explore 171 products in the Inference APIs category. View all Novita AI alternatives.
DeepInfra
Serverless API for 200+ open-weight models on DeepInfra's own US hardware, priced per token.
Inceptron
EU-hosted LLM inference with an in-house compiler: serverless open-weight model APIs and dedicated GPU deployments
LibertAI
Decentralized, privacy-first inference API running open-source LLMs in trusted execution environments
fal
Build the next generation of creativity with fal. Lightning fast inference.
OpenRouter
Unified API for 400+ AI models across 60+ providers, OpenAI SDK-compatible, pay-as-you-go
Work on Novita AI? Feature it at the top of Inference APIs.
Is your product missing?