OpenCode Zen Free Models

OpenCode Zen offers a zero-cost tier for selected coding models to gather community feedback and benchmark performance. These models feature $0.00 input, output, and cached read rates per 1M tokens.

Available Free Models

• Big Pickle: A stealth model available at no cost during its testing period.
• MiMo-V2.5 Free: High-efficiency coding model available free for evaluation.
• Ling 3.0 Flash Fin Free: Specialized flash model provided without token charges.
• Nemotron 3 Ultra Free: NVIDIA-hosted trial model optimized for enterprise-grade reasoning.
• Nemotron 3.5 Lightning Free: Fast-inference NVIDIA trial endpoint.
• Muse Spark 1.3 Contributor Free: Full-capability model offered free in exchange for training contributions.

Access and Setup

1. Create an account on OpenCode Zen and configure billing details to generate an API key.
2. In the OpenCode TUI, run /connect, choose OpenCode Zen, and paste your API key.
3. Access models directly via OpenAI-compatible endpoints or the OpenCode SDK:

• OpenAI-Compatible Endpoint: [https://opencode.ai/zen/v1/chat/completions](https://opencode.ai/zen/v1/chat/completions) (Big Pickle, MiMo, Ling, Nemotron)
• Responses Endpoint: [https://opencode.ai/zen/v1/responses](https://opencode.ai/zen/v1/responses) (Muse Spark)

4. Specify the model in your configuration using the format opencode/<model-id> (e.g., opencode/big-pickle or opencode/nemotron-3.5-lightning-free).

Important Conditions

• Account Requirements: Zen uses a pay-as-you-go architecture requiring valid billing details on file, even when using $0.00 models.
• Data Logging: Free-tier models operate outside Zen's standard zero-retention policy. NVIDIA Nemotron endpoints log session data under API trial terms, and Muse Spark Contributor uses completions for model training. Do not use confidential or sensitive codebases with free-tier endpoints.