Skip to content

Ollama

Ollama is a runtime and cloud service for running open AI models on macOS, Windows, and Linux or through hosted endpoints, aimed at developers and teams building with open models.

Free / Pay-per-useLocal Model Runtimes
230

Overview

Ollama is a runtime and hosted service for working with open AI models, letting developers run them on their own machines or call them through a cloud API. It targets people who want open-model flexibility without locking into a single provider's stack, and it works across macOS, Windows, and Linux.

Local and Cloud Execution in One Toolchain

The core appeal is that the same tooling covers two very different setups. On a local machine, Ollama downloads and runs models directly, which suits laptops and workstations with capable hardware. When a local GPU is not enough, cloud models run on hosted infrastructure, so a project can move between the two without switching tools. The documentation separates local and cloud paths but keeps them under one CLI and API surface.

A Broad Model Library

The model library spans coding, vision, embeddings, and reasoning models, including entries such as gemma4, gpt-oss, kimi, glm, minimax, mistral-large, and nemotron variants. Cloud models use native weights as released by the model provider, and on newer NVIDIA hardware they may use accelerated data formats. Models trained for tool use are tested against real agent workflows before release, according to the vendor.

Developer Integration and Capabilities

Ollama exposes its own API and also works with compatible OpenAI and Anthropic clients, which shortens the path for teams already building against those interfaces. Python and JavaScript libraries are available for application code. Documented capabilities include streaming, thinking, structured outputs, vision, embeddings, speech, tool calling, and web search. A Modelfile reference and CLI reference cover customization and importing existing models, and Docker support is documented for containerized deployments.

Privacy and Data Handling

The vendor states that prompts and responses are never logged or trained on. Cloud models are hosted primarily in the United States, with routing to Europe possible for additional capacity, and hosting partners are required to follow no-logging, no-training, and zero-retention policies. For teams with sensitive workloads, those commitments are the main reason to consider the hosted option rather than a third-party API.

Plans, Credits, and Concurrency

Pay as you go costs nothing to start and bills per token with no subscription or service fees, capped at 3 concurrent requests. Pro adds $60 in monthly usage credits and access to larger pro models. Max raises that to $300 in credits, adds early access to new models, and allows 10 concurrent requests. Team runs $500 per month with unlimited users, $1,000 in shared monthly credits, centralized billing, and priority support. Enterprise pricing is custom and adds model access controls, per-user and per-key cost budgets, and a private support channel. Included credits reset monthly and do not roll over, and requests beyond a plan's concurrency limit are queued.

Where It Fits

Ollama makes the most sense for developers who want a single interface for local experimentation and production cloud calls, and for teams that need shared billing and administrative controls around open models. The per-token model pricing, which varies widely by model and drops during off-peak hours, rewards teams that match model choice to workload rather than defaulting to the largest option.

Core features

  • Runs open models locally on macOS, Windows, and Linux
  • Hosted cloud models for machines without a local GPU
  • API access through Ollama's own API or compatible OpenAI and Anthropic clients
  • Python and JavaScript libraries for application integration
  • Capabilities including streaming, thinking, structured outputs, vision, embeddings, speech, tool calling, and web search
  • Modelfile and CLI references for customizing and importing models
  • Docker support and hardware guidance for GPU setups
  • Team and enterprise administration with centralized billing and model access controls

Use cases

  1. 1Running coding agents and reasoning models from a local terminal
  2. 2Connecting desktop apps such as Claude or ChatGPT to open models
  3. 3Building applications that call cloud models through OpenAI- or Anthropic-compatible APIs
  4. 4Prototyping with vision, embedding, or speech models from the model library
  5. 5Giving a team shared usage credits and centralized billing
  6. 6Enforcing model access limits and cost budgets across an organization

Strengths and things to check

Strengths

  • Supports both local execution and hosted cloud models under one toolchain
  • Prompts and responses are not logged or used for training, per the vendor
  • Broad model library spanning coding, vision, embeddings, and reasoning
  • Compatible with OpenAI and Anthropic client libraries, which lowers integration effort
  • Free pay-as-you-go tier with no subscription and no service fees

Things to check

  • Cloud usage is billed per token, so costs scale with volume and model choice
  • Concurrency is capped at 3 requests on Pay as you go and Pro, and 10 on Max and Team
  • Included monthly credits do not roll over to the next month
  • Only one account is allowed per person
  • Hosting is primarily in the United States, with some routing to Europe for capacity

Pricing

Pay as you go is $0 with per-token billing, no subscription, no service fees, and 3 concurrent requests. Pro adds $60 of monthly usage credits and access to larger pro models. Max adds $300 of monthly credits, early access to new models, and 10 concurrent requests. Team is $500 per month with unlimited users, $1,000 of shared monthly credits, centralized billing, and priority support. Enterprise is custom-priced with model access controls, cost budgets, and dedicated support. Cloud models are billed per million tokens, with off-peak rates outside 12:00 and 18:00 UTC on weekdays and all day on weekends. Prices are subject to change.

Basic information

Website
ollama.com
Added
Oct 8, 2026
Platforms
Web
Chinese & English
Not marked
API
Not marked
Self-hosting
Not marked
Best for
Ollama suits developers, data teams, and organizations that want to run open models on their own hardware or through a hosted API while keeping prompts and responses private.
Information updated
Oct 16, 2026
Website status
Checking
Corrections · Privacy · Data requests?
Contact us

Features, pricing and availability may change. Always verify details on the official website. Maintained by steven · Last updated Oct 16, 2026 →

More on this tool

Related tools

All categories

LlamaIndex

Enterprise Grade

LlamaIndex builds document parsing and extraction software plus open-source tools for AI applications, helping teams turn enterprise documents into usable data for AI agents and workflows.

Free TierSOC 2
View details