OpenCode Go
$10/mo
What you get
- Coding agent
- Works with OpenCode and other validated agents (e.g. Claude Code, Codex)
- Included usage
- Per-model $ limits; e.g. ~110 Kimi K3 or ~3,850 Claude Haiku 5.5 requests per 5h (estimates)
- Limit windows
- 5-hour (20%), weekly (50%), monthly (100%)
- If you run out
- Optional pay-as-you-go balance fallback; free models remain usable
Quota details
- Limits are per-model monthly dollar amounts; 5-hour = 20%, weekly = 50%, monthly = 100% of that amount.
- Estimates assume typical agent requests (e.g. ~55k cached input tokens per request); actual usage varies.
- Single-member workspaces only (no team invites).
- Some 'Contributor' models (Gemini 4 Argon Contributor, Muse Spark) allow the provider to train on your prompts; others are not used for training.
- OpenCode says usage limits may change; some model limits shown are limited-time boosts.
Published estimates (requests per 5 hours, by model)
| Model | Est. requests / 5h | Monthly $ limit |
|---|---|---|
| Kimi K3 | 110 | $15 |
| GLM-5.2 | 880 | $60 |
| Kimi K2.7 Code | 1,350 | $60 |
| Claude Haiku 5.5 | 3,850 | $15 |
| GPT 6 Luna | 4,230 | $15 |
| Grok 4.7 | 169 | $15 |
| MiniMax M3 | 3,200 | $60 |
| DeepSeek V4.1 Flash | 34,670 | $80 |
Caveats
- Frontier closed models (Claude Opus/Sonnet, GPT Sol/Astra) are not in the Go lineup; it is open/budget models plus Haiku/Luna-class models.
Not published / unknown
We don't guess. These are the things we could not confirm from a source.
- Model list changes over time
Community burn reports
No burn reports for this plan yet.
Be the first: tell us how fast you hit your limits.
Report your burn on OpenCode Go
Be the first: tell us how fast you hit your limits.
Report your burn on OpenCode Go
Changes affecting this plan
- No logged changes yet.
Sources
Every figure on this page comes from these pages, read on the dates shown.