OpenCode TUI plugin that shows token usage, context window, generation speed, cost, and file changes in the sidebar after each response.
opencode plugin opencode-stats-for-nerdsRestart OpenCode. A Stats for Nerds panel appears in the right sidebar.
▼ Stats for Nerds
Tokens 13.4k in · 1.2k out · 2.0k thinking
Cached 3.8k read · 2.7k write
Context 23.2k / 200k (11%)
────────────────
Cost $0.0070
Gen Time 2m 14s
TTFT 1.35s
Speed 42.3 tok/s
────────────────
Changes +234 -56 · 5 files
Model claude-sonnet-4
Click the header to collapse/expand.
| Stat | Scope | Description |
|---|---|---|
| Tokens | Cumulative | Input, output, and thinking tokens across all turns |
| Cached | Cumulative | Prompt cache hits (read = reused, write = newly cached) |
| Context | Current | Context window used vs model limit, with percentage |
| Cost | Cumulative | Total session cost at provider rates |
| Gen Time | Cumulative | Total time the model spent generating |
| Think Time | Cumulative | Total time the model spent reasoning (off by default) |
| TTFT | Last response | Time-to-first-token |
| Speed | Last response | Output tokens per second |
| Activity | Cumulative | Tool-call steps and invocations (off by default) |
| Changes | Session | File additions, deletions, and count |
| Model | Last response | Model ID |
Disable any stat via tui.json:
{
"$schema": "https://opencode.ai/tui.json",
"plugin": [
["opencode-stats-for-nerds", {
"show": {
"cache": false,
"activity": true,
"model": false
}
}]
]
}All options default to true except thinkTime and activity.
| Key | Default | Controls |
|---|---|---|
tokens |
true |
Input/output/thinking row |
cache |
true |
Cache read/write row |
context |
true |
Context window used / limit (%) |
cost |
true |
Session cost |
genTime |
true |
Total generation time |
thinkTime |
false |
Total reasoning time |
ttft |
true |
Time-to-first-token |
speed |
true |
Tokens per second |
activity |
false |
Steps and tool calls |
changes |
true |
File additions/deletions |
model |
true |
Model name |
- OpenCode v1.17+
MIT