Models & credits

FiClaw models and credits

FiClaw routes cloud models through FiClawCloud and records model usage with credits. This page explains model choice, how credits are charged, and what the membership center shows.

Current models

Current cloud model directory

The model directory is maintained by FiClawCloud; the desktop app enables the model set after launch and login.

Claude

claude-haikuclaude-sonnetclaude-opus

GPT

gpt-5.5gpt-5.4gpt-5-nanogpt-5-mini

Gemini

gemini-3-flash-previewgemini-3.1-pro-preview

DeepSeek

deepseek-v4-flashdeepseek-v4-pro

Other models

qwen-3.6-27bkimi-k2.6glm-5.1minimax-m2.7

Credits

How credits are consumed

01

Credits are consumed per model call

FiClawCloud computes credit usage from actual input tokens, output tokens, the model rate and a minimum charge.

02

Different models have different rates

Lightweight models suit high-frequency everyday tasks; higher-capability models suit complex research, long-context analysis and high-quality reasoning.

03

The membership center shows the current basis

The desktop membership center shows balance, top-up history, usage history, model rate notes and typical per-round multipliers.

04

Pilots can confirm a budget boundary first

If a team has fixed scenarios, concurrent usage or budget requirements, book a demo first to confirm model choice and pilot scope.

Model choice

How to choose a model

Everyday tidying and light Q&A

Prefer fast, low-cost models for material organization, summaries, format conversion and low-risk assistance.

Research analysis and complex reasoning

Prefer higher-capability models for long-document analysis, strategy-logic derivation, review summaries and multi-step judgment.

Long-context material processing

When a task includes long reports, history or multi-turn context, a long-context model fits better.

Pilots and cost control

Set a model mix and credit budget around one high-value scenario, then expand to more roles and workflows.