aichat Review, Pricing & Alternatives (2026)

aichat is a free, open-source Rust terminal client maintained in the sigoden/aichat repository, created on March 3, 2023. It targets developers who want one command-line tool for chat, shell assistance, retrieval-augmented generation, agents, and local services, rather than another hosted model subscription. The project offers installation through Cargo, Homebrew, Pacman, Scoop, Termux, and prebuilt releases, supporting macOS, Linux, and Windows. Its dual MIT or Apache License 2.0 terms permit broad reuse, while its provider-neutral design lets users connect cloud services or local systems such as Ollama and OpenAI-compatible endpoints. aichat therefore operates as client software and can run a local server, not as an inference vendor. Its repository had more than 10,400 GitHub stars in a 2026 review and continued receiving updates, including release 0.30.0.

The typical workflow starts in a chat REPL, a one-shot command, or a shell assistant that converts natural-language requests into commands. Users can attach files, directories, images, URLs, or standard input, then ask a configured model to summarize, transform, or answer questions. Built-in RAG indexes documents for grounded responses. Agents and AI tools call configured capabilities, while roles, macros, sessions, editor integration, and command execution extend the REPL. Running `aichat --serve` starts a local HTTP service with OpenAI-compatible chat-completions, embeddings, and reranking endpoints, along with playground and arena interfaces. The client supports cloud and local providers, including OpenAI, Anthropic, Gemini, Ollama, Groq, Azure OpenAI, Vertex AI, Bedrock, and arbitrary OpenAI-compatible endpoints. This design makes aichat useful for scripts, internal applications, and developer workflows.

aichat is built around replaceable provider integrations rather than a proprietary model stack. It supports more than 20 provider types, including OpenAI, Claude, Gemini, Ollama, Groq, Azure OpenAI, Vertex AI, Bedrock, GitHub Models, Mistral, DeepSeek, xAI, Cohere, Perplexity, Cloudflare, OpenRouter, Qianwen, Moonshot, ZhipuAI, MiniMax, DeepInfra, and VoyageAI. Arbitrary OpenAI-compatible endpoints extend that list for compatible hosted or local services. The model catalog includes GPT-5.2, GPT-5, Gemini 2.5, Gemini 3 preview, and Claude Opus 4.6. Release 0.30.0 added OSC52 clipboard support, automatic dark and light theme detection, REPL improvements, and tool and agent fixes. Provider credentials can be configured through variables such as OPENAI_API_KEY and GEMINI_API_KEY.

aichat itself costs $0 and includes no hosted subscription or bundled inference allowance. Its MIT or Apache 2.0 license covers the client, while configured providers charge separately for model requests. OpenAI, Anthropic, Google, AWS, Groq, and other services set their own token, request, model, and account limits. Users can avoid remote inference charges by connecting a local service such as Ollama, although local models require suitable hardware and a running service. API keys, including OPENAI_API_KEY and GEMINI_API_KEY, can be set through environment variables or configuration files. The built-in server can bind to a chosen address and port, making network exposure an operator decision. There is no single monthly bill covering the client, provider usage, infrastructure, and support. Costs therefore depend on the selected model, request volume, and deployment arrangement.

Flexibility brings setup work and operational responsibility. aichat does not select a model, provide a hosted privacy boundary, offer one support desk, or impose a predictable monthly inference bill. The $0 client is MIT or Apache 2.0 licensed, but cloud requests still use credentials and incur provider charges, while Ollama can avoid remote inference fees if suitable hardware and a running service are available. Shell execution requires permission controls, and its local HTTP server still needs protection when exposed beyond the machine. Aider focuses more tightly on repository editing, whereas aichat combines chat, shell, RAG, agents, server functions, and more than 20 provider types. ShellGPT offers a similar command-line workflow with OpenAI and Ollama support, while aichat adds RAG and agents. Claude Code is a more integrated single-provider coding environment. Pick aichat if you want a licensed, provider-neutral terminal layer over cloud or local models.