Free Chat Models

The fastest, cheapest GPT-5 version, suited to summarization and classification. 400K context window with text and image input.

More than twice as fast as GPT-5 mini, with better coding, reasoning, multimodal understanding, and tool use. Meant for responsive coding assistants, quick subagents, and screenshot-reading systems.

The smallest, cheapest GPT-5.4 model, recommended for classification, data extraction, ranking, and subagents supporting larger systems.

Released in 2024 as a cheap small model with a 128K context window and text and vision input. Older, but fine for simple chat and summaries.

An open-weight reasoning model under the Apache 2.0 license, with adjustable reasoning effort and a 131K context. Matches or beats o3-mini on OpenAI's benchmarks despite its small size.

The larger open-weight sibling, with near-parity with o4-mini on core reasoning benchmarks while running on a single 80 GB GPU. The stronger pick for harder reasoning and tool use.

Low-latency, cost-efficient model for high-volume work. Up to 1M input tokens and text, image, audio, video, and PDF input, with adjustable thinking levels.

Tuned for simple coding, precise document understanding, and lightweight agent workflows at minimal cost. Google calls it its fastest, most cost-effective 3.5-class model.

Reasoning model with a 1M-token context window and a December 2025 knowledge cutoff. Supports tool use, vision, file input, and web search. Targets agentic workflows, instruction following, and factual accuracy.

Open model with 550B total and 55B active parameters, a 1M-token context, and a design aimed at long-running agent workflows. Independent testing ranked it highest in intelligence among open-weight models from US developers.

109B-parameter model with a 10M-token context window that fits on a single H100 GPU. Best for huge documents and large codebases.

400B-parameter generalist with a 1M-token context window, aimed at GPT-4o and Gemini 2.0 Flash. Handles text and images, so it suits assistants and chat.

Updated 24B model with better instruction following, fewer repetitive outputs, and stronger function calling. Image and text input, 128K context.

Lightweight search model built on Llama 3.3 70B, tuned for speed and quick factual queries. Combines a language model with web search and citations, so use it when you want sourced answers.

Dialogue-first model for immersive roleplay, character-driven chat, and multi-turn conversation. Text-only and a specialist, not a general-purpose replacement for MiniMax's other models.

Small, fast model that excels at RAG, tool use, agents, and multi-step reasoning. 128K context window.

No page found under this name. The Nov 2023 date matches the launch of OpenRouter's Auto Router, which picks a model for each prompt, so this is probably a label for it. This is an inference, so check it in the app.