Nemotron 3 Ultra API model, prices & limits

Developer: NVIDIA · API model

Canonical model entity with no single official token price; NVIDIA publishes this base model and the actual call price, currency and conditions live on each channel offer.

Catalog information only; CreditsPlan does not run inference with this model.

API models and prices · 1

Usage-based billing: text rates are usually per million tokens; other units and tiers are stated per entry. Compare only the same currency, unit and billing conditions. Unverified cache pricing does not mean free.

Nemotron 3 Ultra

Pricing Not Public

Endpoint billing is unverified; excluded from lowest-cost comparisons.

Type / status: llm / documented_api

Use cases: Model price comparison, pay-as-you-go API selection, cross-provider call cost comparison

Excluded from default production recommendations; check model availability, pricing and usage conditions.

No single official NVIDIA token price: this base model is sold by several providers at different rates. Compare the reviewed channel offers for the actual per-provider price, currency and conditions.

Official documentation

Official model ID: nemotron-3-ultra

Model developer: NVIDIA · Service provider: NVIDIA · Market: com

Model family: Nemotron · Model catalog only; see service-channel quotes

Canonical model entity with no single official token price; NVIDIA publishes this base model and the actual call price, currency and conditions live on each channel offer.

Documented API context: 1,000,000 tokens

Official source · Last verified: 2026-11-01T00:00:00.000Z

Service channels / prices and conditions 2

Confirmed token quotes use USD / 1M tokens. Other billing units remain available at the source. Each channel retains its own price verification status; catalog retrieval dates are not price verification dates.

DigitalOcean Inference Model developer

Region: Global listing · Usage-based API · Metered billing · USD

Input $0.9 / 1M tokens · Output $1.7 / 1M tokens · Cached read $0.18 · Cached write Not disclosed

nemotron-3-ultra · NVIDIA Nemotron 3 Ultra / nemotron-3-ultra / nemotron-3-ultra

DigitalOcean-hosted model rates. Serverless inference is prepaid only: promotional credits or a positive prepaid balance are required and access is suspended when both run out. Commercial Anthropic and OpenAI models billed by the original provider when using your own API keys. Batch inference discounts OpenAI and Anthropic models by up to 50%. Dedicated inference is billed per GPU-hour (for example H100 at $4.41/hour) and is not a model API price.

Cache billed separately

Channel context / maximum output: Unverified / Unverified tokens

API: https://inference.do-ai.run/v1 · openai-compatible · api_key

Tools: Supported · Image input: Unverified

First-party price verified · 2026-11-01T00:00:00.000Z

Official documentation · Data source

Ollama Cloud Model developer

Region: Global listing · Usage-based API · Metered billing · USD

Input Not disclosed / 1M tokens · Output $3 / 1M tokens · Cached read $0.1 · Cached write Not disclosed

nemotron-3-ultra · Nemotron 3 Ultra / nemotron-3-ultra / nemotron-3-ultra

Ollama Cloud bills per token and requires purchased usage credits; concurrency is 3 (Pay as you go and Pro) or 10 (Max and Team). Off-peak rates (outside 12:00–18:00 UTC on weekdays and all weekend) are published separately by Ollama and are not folded into the standard rate shown here.

Cache billed separately

Channel context / maximum output: Unverified / Unverified tokens

Data use: Ollama states prompts and responses are never logged or trained on; models and compute are hosted primarily with NVIDIA Cloud Providers under no-logging, no-training and zero-retention requirements.

API: https://ollama.com/v1 · openai-compatible · api_key

Tools: Supported · Image input: Unverified

First-party price verified · 2026-11-01T00:00:00.000Z

Official documentation · Data source

Input means uncached tokens; cache read and write are added separately. Context tiers require the context size of each request, not monthly totals. Estimates exclude reasoning, audio and unconfirmed charges; top-ups are account funding, not additional token consumption.

服务渠道 / 价格与条件 2

中英文报价均使用美元 / 百万 tokens。各渠道分别保留来源和价格核验状态；目录抓取时间不等于价格核验时间。

DigitalOcean Inference 原厂

地区：国际区 · 按量 API · 按量计费 · USD

输入 $0.9 / 1M tokens · 输出 $1.7 / 1M tokens · 缓存读 $0.18 · 缓存写未公布

nemotron-3-ultra · NVIDIA Nemotron 3 Ultra / nemotron-3-ultra / nemotron-3-ultra

DigitalOcean 自营托管模型价格。Serverless inference 仅支持预付：需促销 credits 或正数预付余额，用尽后暂停访问。Anthropic / OpenAI 等商业模型在使用自带 API Key 时由原厂直接计费。Batch inference 对 OpenAI 与 Anthropic 模型最高 5 折。Dedicated inference 按 GPU 小时另计（如 H100 $4.41/小时），不是模型 API 价格。

缓存可能另计

此渠道上下文 / 最大输出：待核验 / 待核验 tokens

API: https://inference.do-ai.run/v1 · openai-compatible · api_key

工具调用: 支持 · 图像输入: 待核实

一手价格已核对 · 2026-11-01T00:00:00.000Z

官方文档 · 数据来源

Ollama Cloud 原厂

地区：国际区 · 按量 API · 按量计费 · USD

输入未公布 / 1M tokens · 输出 $3 / 1M tokens · 缓存读 $0.1 · 缓存写未公布