AI Free Member Models

June 2026

The following free models are available through the Gnoppix AI API. Each model includes its vendor and primary use case.

Model List

`nex-agi/nex-n2-pro:free`

Vendor: Nex AGI
Use case: Agentic mixture-of-experts model (397B total, 17B active) built on Qwen3.5. Designed for coding, tool use, deep research, and long-horizon agentic workflows. Supports reasoning, function calling, and structured outputs. Accepts text and image input.

`openrouter/owl-alpha`

Vendor: OpenRouter
Use case: High-performance foundation model designed for agentic workloads. Natively supports tool use and long-context tasks (1M context). Strong in code generation, automated workflows, and complex instruction execution. Compatible with Claude Code and other productivity tools.

`nvidia/nemotron-3.5-content-safety:free`

Vendor: NVIDIA
Use case: Content safety moderator (4B parameters, fine-tuned Gemma-3-4B-it). Evaluates prompts, images, and responses for safety. Supports custom policy enforcement with reasoning traces, multilingual moderation (12 languages), and multimodal inputs.

`nvidia/nemotron-3.5-ultra-550b-a55b:free`

Vendor: NVIDIA
Use case: Large-scale reasoning model (550B total, 55B active). Designed for complex reasoning, analysis, and high-quality text generation across enterprise workloads.

`nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free`

Vendor: NVIDIA
Use case: Lightweight reasoning model (30B total, 3B active). Optimized for efficient inference with strong reasoning capabilities for resource-constrained environments.

`nvidia/nemotron-3-nano-30b-a3b:free`

Vendor: NVIDIA
Use case: General-purpose nano model (30B total, 3B active). Compact MoE model for efficient text generation and instruction following.

`nvidia/nemotron-3-super-120b-a12b:free`

Vendor: NVIDIA
Use case: Large MoE model (120B total, 12B active). Designed for high-quality text generation, reasoning, and complex task completion.

`nvidia/nemotron-nano-12b-v2-vl:free`

Vendor: NVIDIA
Use case: Vision-language model (12B). Processes both text and images for multimodal understanding tasks.

`nvidia/nemotron-nano-9b-v2:free`

Vendor: NVIDIA
Use case: Compact general-purpose model (9B). Efficient text generation and instruction following for lightweight deployments.

`poolside/laguna-xs.2:free`

Vendor: Poolside
Use case: Efficient coding agent model (33B total, 3B active MoE). Second-generation open-weight model under Apache 2.0. Designed for agentic coding workflows with tool calling and reasoning. Runs on a single GPU.

`poolside/laguna-m.1:free`

Vendor: Poolside
Use case: Flagship coding agent model (225B total, 23B active MoE). Optimized for complex software engineering tasks. Supports tool calling and reasoning with 128K context. Quantized to fp8 for efficient inference.

`google/gemma-4-26b-a4b-it:free`

Vendor: Google
Use case: Instruction-tuned model (26B total, 4B active). Lightweight MoE model for general-purpose chat, instruction following, and text generation tasks.

`google/gemma-4-31b-it:free`

Vendor: Google
Use case: Instruction-tuned model (31B dense). General-purpose chat and instruction following with strong reasoning capabilities.

`liquid/lfm-2.5-1.2b-thinking:free`

Vendor: Liquid AI
Use case: On-device reasoning model (1.2B). Optimized for math, logic, and multi-step problem-solving with chain-of-thought. Runs under 1GB memory — ideal for edge deployment.

`liquid/lfm-2.5-1.2b-instruct:free`

Vendor: Liquid AI
Use case: Instruction-tuned model (1.2B). Designed for chat, instruction following, and tool calling on edge devices. Fast inference on CPU and mobile NPU.

`qwen/qwen3-next-80b-a3b-instruct:free`

Vendor: Alibaba Cloud (Qwen team)
Use case: Next-generation instruction model (80B total, 3B active MoE). General-purpose chat and instruction following with efficient inference.

`qwen/qwen3-coder:free`

Vendor: Alibaba Cloud (Qwen team)
Use case: Code-specialized model. Designed for code generation, debugging, and software engineering tasks.

`openai/gpt-oss-120b:free`

Vendor: OpenAI
Use case: Open-weight reasoning model (117B total, 5.1B active MoE, Apache 2.0). Strong reasoning, tool use, and agentic capabilities. Fits into a single H100 GPU. Configurable reasoning effort.

`openai/gpt-oss-20b:free`

Vendor: OpenAI
Use case: Compact open-weight reasoning model (21B total, 3.6B active MoE, Apache 2.0). Runs within 16GB memory — ideal for local deployment and edge devices. Configurable reasoning effort.

`meta-llama/llama-3.3-70b-instruct:free`

Vendor: Meta
Use case: Large instruction-tuned model (70B). General-purpose chat, reasoning, and text generation. One of the most widely adopted open-source LLMs.

`meta-llama/llama-3.2-3b-instruct:free`

Vendor: Meta
Use case: Small instruction-tuned model (3B). Efficient text generation and chat for lightweight and on-device use cases.

`nousresearch/hermes-3-llama-3.1-405b:free`

Vendor: Nous Research
Use case: Very large instruction-tuned model (405B). Built on Llama-3.1-405B, fine-tuned for high-quality reasoning, instruction following, and complex task completion.