Best fit
Implementation on any backend you choose, and zero-cost baselines in comparison runs.
OpenCode is an open-source terminal agent that runs on many model backends, free ones included. Each task gets the same planning, isolated workspace and GitHub PR loop as every other agent.
OpenCode is open source and free. Its built-in free models, Big Pickle first, cost nothing for a limited time. Go at $10 a month and Go Plus at $40 add 32 curated paid models with a monthly dollar allowance per model, counted over five-hour, weekly and monthly windows. OpenCode publishes those dollar figures and no request count for the free models ProPR lists.
| Plan | Price (USD) | Models | Published limits |
|---|---|---|---|
| Free | $0 | Big Pickle, Beam Free, DeepSeek V4.1 Flash Free, Flint 40B Free, GLM 5.3 Flash Free, MiMo V2.6 Flash Free, Nemotron 3.5 Lightning and Super VL Free. | Daily request limit on free models. No number published for these models. |
| OpenCode Go | $10 / month | 32 curated models: GLM-5.3-Flash, DeepSeek V4.1 Flash, Kimi K3, MiMo-V2.6-Flash, GPT-6 Luna, Claude Haiku 5.5 and more. | Monthly allowance per model: $60 GLM-5.3-Flash, $100 DeepSeek V4.1 Flash, $15 Kimi K3 and GPT-6 Luna. Extra Usage draws on credit. |
| OpenCode Go Plus | $40 / month | Same 32 models. | Higher allowances: $180 GLM-5.3-Flash, $200 DeepSeek V4.1 Flash, $60 Kimi K3 and GPT-6 Luna. |
| Console credits or your own key | Pay-as-you-go per token | Every Console model, plus any provider key you authenticate. | Metered billing with spend budgets; no plan cap. |
Every option runs headless: opencode run runs one task and exits, which is how ProPR drives it.
Sources: OpenCode Go, OpenCode Zen model catalog, OpenCode CLI docs. Verified November 2026.
That plan runs in ProPR on GitHub issues and pull requests. Label an issue llm-opencode-big-pickle or llm-opencode-glm53-flash-free and ProPR plans the task, runs OpenCode in an isolated workspace, and opens the PR. You pay OpenCode or your provider directly; ProPR adds no token markup. Read how an agent subscription becomes pull requests from GitHub issues, OpenCode included.
Implementation on any backend you choose, and zero-cost baselines in comparison runs.
Built-in free models, an OpenCode plan, or any provider key you already have.
OpenCode runs implementation, review and follow-up passes on whichever backend you authenticate.
ProPR's OpenCode lineup starts with free models. Use them for low-stakes passes and zero-cost baselines in multi-model comparison runs, and add OpenCode's paid models (DeepSeek, MiMo, MiniMax, GLM, Kimi, Qwen) when you want them.
| Model | Issue label | Context | Notes |
|---|---|---|---|
| Big Pickle (default) | llm-opencode-big-pickle | 200K | OpenCode's own model and ProPR's default for everyday tasks. |
| Beam Free | llm-opencode-beam-free | 1M | Free passes over a large codebase. |
| DeepSeek V4.1 Flash Free | llm-opencode-deepseek-v41-flash-free | 200K | Fast coding passes and quick follow-ups. |
| Flint 40B Free | llm-opencode-flint-40b-free | 262K | A small open-weight model for low-stakes passes. |
| GLM 5.3 Flash Free | llm-opencode-glm53-flash-free | 200K | A free baseline for comparison runs. |
| MiMo V2.6 Flash Free | llm-opencode-mimo-v26-flash-free | 200K | Another free baseline for comparison runs. |
| Nemotron 3.5 Lightning Free | llm-opencode-nemotron35-lightning-free | 262K | Quick, low-stakes passes at zero token cost. |
| Nemotron 3.5 Super VL Free | llm-opencode-nemotron35-super-vl-free | 1M | Free passes that need a large context window. |
OpenCode setup, models and credentials in the ProPR docs → ProPR runs the official OpenCode CLI.