Use GLM-5.3, Kimi K3, DeepSeek V4.1 Flash, Muse Spark 1.3, Qwen3.8, and more through your ClinePass subscription — directly inside pi with full streaming, tool calls, and thinking level support.
pi install npm:pi-clinepass-provider
Requires a ClinePass subscription — $9.99/month
12 curated models with per-model thinking level support
| Model | Context | Input | Output | Reasoning |
|---|---|---|---|---|
| GLM-5.3 cline-pass/glm-5.3 | 1M | $1.40/M tok | $4.40/M tok | low / high / xhigh → max |
| GLM-5.3-Flash cline-pass/glm-5.3-flash | 1M | $0.15/M tok | $0.50/M tok | low / high / xhigh → max |
| Kimi K3 cline-pass/kimi-k3 | 1M | $3.00/M tok | $15.00/M tok | high → max (always on) |
| Muse Spark 1.3 Contributor cline-pass/muse-spark-1.3-contributor | 1M | $0.10/M tok | $0.20/M tok | minimal → xhigh (always on) |
| DeepSeek V4 Pro cline-pass/deepseek-v4-pro | 1M | $1.74/M tok | $3.48/M tok | off / high |
| DeepSeek V4.1 Flash cline-pass/deepseek-v4.1-flash | 1M | $0.30/M tok | $1.20/M tok | off / high |
| MiMo-V2.5 cline-pass/mimo-v2.5 | 262K | $0.14/M tok | $0.28/M tok | off → high |
| MiMo-V2.5-Pro cline-pass/mimo-v2.5-pro | 262K | $1.74/M tok | $3.48/M tok | off → high |
| MiniMax M3 cline-pass/minimax-m3 | 1M | $0.30/M tok | $1.20/M tok | off → high |
| Qwen3.7 Max cline-pass/qwen3.7-max | 262K | $2.50/M tok | $7.50/M tok | off → high |
| Qwen3.7 Plus cline-pass/qwen3.7-plus | 1M | $0.40/M tok | $1.60/M tok | off → high |
| Qwen3.8 Max cline-pass/qwen3.8-max | 1M | $2.00/M tok | $6.00/M tok | low / medium / xhigh |
Prices are usage estimates, not bills or exact ClinePass quota measurements. The three new models use upstream reference rates, not confirmed ClinePass rates. DeepSeek V4.1 Flash uses peak rates; upstream off-peak rates are half.
Muse Spark Contributor: Meta's Contributor tier permits training on prompts and completions. Check Cline's applicable terms before sending private code. Muse Spark is hosted, not open-weight.
All models are text-only in this extension. Reasoning labels use pi levels: GLM-5.3 and Flash support low/high/xhigh (xhigh sends max), with thinking always on.
This update removes the old IDs immediately ahead of Cline's announced September 21 cutoff. Migrate GLM-5.2 to GLM-5.3, Kimi K2.6/K2.7 Code to Kimi K3, and DeepSeek V4 Flash to V4.1 Flash. No aliases are provided.
Everything you need to use ClinePass models directly in pi
Native OpenAI-compatible streaming via pi's openai-completions — SSE parsing, tool calls, and usage tracking out of the box.
Map pi's six thinking levels to each model's reasoning_effort. Set with --thinking and switch in-session with /thinking.
Reuses your existing Cline CLI login — no separate API key needed. Tokens refresh automatically behind the scenes.
Fetches live model list from the Cline API at startup. Falls back to a curated static catalog if the endpoint is unavailable.
Per-token usage tracking against ClinePass reference pricing so you know exactly what each model session costs.
Auto-detects existing WorkOS credentials or falls back to browser-assisted manual API key paste in one command.
Two paths — pick the one that fits
If you already use the Cline CLI and have run cline auth, this extension detects your WorkOS OAuth credentials automatically. Run pi /login and select ClinePass — no separate API key needed. Short-lived tokens refresh server-side.
Generate an API key from app.cline.bot → Settings → API Keys. Set CLINE_API_KEY in your environment, or paste it during the pi /login flow. Keys never expire.