OpenRouter Free AI Models

OpenRouter offers a rotating collection of zero-cost models accessible via a unified OpenAI-compatible API. These models are hosted to support developer experimentation, prototyping, and benchmarking without requiring upfront compute spend.

Key Model Categories Included

• Coding & Software Engineering Agents: Includes specialized models such as Poolside's Laguna S 2.1 and Laguna XS 2.1, Cohere's North Mini Code, and Nex AGI's Nex-N2.5 series designed for autonomous software development and terminal tool-use workflows.
• General Reasoning & MoE: Large-context reasoning models such as Thinking Machines' Inkling and Inkling Small (up to 1.05M tokens context), NVIDIA Nemotron 3.5 Lightning (1M context), and Dots Studio's Dots3-Note Preview.
• Domain-Specific Models: InclusionAI’s Ling 3.0 Flash Fin (finance workflows) and Ling 3.0 Flash Sante (medical reasoning and clinical safety).
• Audio & Speech Synthesis: Text-to-speech options including Deepgram Flux TTS and Fish Audio S2.1 Pro Free.
• Embeddings & Reranking: NVIDIA Nemotron 3 Embed 1B, LiquidAI LFM2.5-Embedding-350M, and multimodal document reranker NVIDIA Llama Nemotron Rerank VL 1B V2.

Usage Considerations

Free variants on OpenRouter operate under standard rate limits and shared capacity. Depending on provider terms (such as LiquidAI or Poolside), prompts and outputs submitted to free variants may be logged or retained by the respective model creators for research and training purposes.