AI-assisted search over the gt_server project's chat modules and content directory.

Local 120-billion-parameter open-source LLM with a 120K-token context window, served by the on-prem 本地服務.

Meta's Llama 4 MoE model (16 experts x 17B active) with a 120K context window, running locally via local server.

Google Gemini 3.5 Flash Lite, the fastest and most affordable tier of the Gemini 3.5 family.

Google Gemini 3.6 Flash, the newest fast general-purpose Gemini model at a lower cost.

Google Gemini 3.1 Flash Lite, the fastest and most affordable Gemini tier.

Google Gemini 3.5 Flash, the latest fast and cost-efficient general-purpose Gemini model.

Google Gemini 3.1 Pro, the flagship Gemini model via Google AI Studio.

Anthropic Claude Haiku 4.5, the fastest and cheapest Claude model.

Anthropic Claude Sonnet 5, a balanced mid-tier model with strong reasoning and writing quality.

Anthropic Claude Opus 4.8, the flagship of the Claude 4 family, with sharper judgment and improved honesty.

Anthropic Claude Opus 5, the new-generation flagship with a large jump in coding, reasoning, and computer use.

Anthropic Claude Fable 5, the most powerful Claude model, a new tier above Opus.

OpenAI GPT-5.6 Luna — built for cost-sensitive, high-volume workloads.

OpenAI GPT-5.6 Terra — balances intelligence and cost for general-purpose tasks.

OpenAI GPT-5.6 Sol — frontier flagship for complex professional work, 1.05M context and 128K max output.

OpenAI ChatGPT 5.4 nano, the smallest and fastest 5.4-family model.

OpenAI ChatGPT 5.4 mini, a lighter and cheaper variant of GPT-5.4.

OpenAI ChatGPT 5.5 — flagship with 1.05M context and 128K max output.

Perplexity Sonar Pro, a web-search-augmented LLM.

Perplexity Sonar Reasoning Pro combines live web search with deeper reasoning.

Perplexity Sonar Deep Research runs a multi-step research workflow: planning, iterative searching, cross-referencing, and synthesis.

xAI Grok 4.5, flagship model with strong reasoning and a 500K context window.

BytePlus Seed 2.0 Lite, the lightweight Seed tier.

BytePlus Seed 1.8, a strong Chinese-optimized LLM.

MoonshotAI Kimi K3 flagship model with vision and a 1M-token context.

Z.ai GLM 5.2 large-scale reasoning model with a 1M-token context.

Alibaba Qwen 3.5 MoE (397B total, 17B active).

Upload documents or paste text, and the AI splits the content into knowledge blocks by chapter, paragraph, or concept, showing the result in JSONL format and producing a downloadable JSONL file.

Site Default Model

Paste a meeting transcript (from speech_to_text or manual notes) and receive a clean summary with key decisions, action items, and follow-up dates.

One-click slide outline generator.

Generate a 20-slide pptx deck via gamma.app.

Card-based notebook manager for organising reading, research, and project notes.

An AI copy-editor.

Generates Taiwan-style official documents (公文) following the standard format and phrasing conventions.

Upload a PDF and ask questions about its contents.

An AI homework assistant that explains solutions step by step.

Takes your draft social-media post and rewrites it for higher engagement.

Contract review assistant.

Structured report writer.

Paste a URL and receive a concise summary of the page's key points.

Enter keywords to find related web pages via Google Search — returns titles, snippets and clickable URLs.

Enter keywords to find related news via Google News — returns titles, sources, publish times and clickable links.

Academic paper assistant.

Cloud speech recognition powered by OpenAI's gpt-4o-transcribe model, with plain-text transcript and SRT subtitle output.

On-premise Whisper Large v3 speech recognition with plain-text transcript and SRT subtitle output.

Single-speaker local speech-to-text. Outputs plain text and SRT subtitles without diarization.

Multi-speaker local speech-to-text with speaker diarization. Tags each segment with SPEAKER_01 / SPEAKER_02 etc.

Cloud multi-speaker speech recognition powered by OpenAI's gpt-4o-transcribe-diarize model, with plain-text transcript and SRT subtitle output.

Programming assistant that answers coding questions, explains errors, reviews snippets, and suggests refactors.

Project plan generator.

Rewrites event material into formal written press-release style, suitable for media outreach, internal newsletters, and PR distribution.

Adapts news material into a spoken-word broadcast script — short sentences, natural rhythm, easy to read aloud.

Upload an image or screenshot; AI extracts editable plain text via OCR.

Paste any article and receive an AI-vs-human authorship likelihood score with per-paragraph breakdown.

Long-form summarizer that distills your input down to approximately 500 Chinese characters (or equivalent English).

Detailed summarizer targeting ~1,000 Chinese characters.

General-purpose writing optimizer.

AI essay grader that assigns A / B / C / D / E with specific feedback on structure, content, and language.

Enter a URL to crawl the page and organize it into title, description, keywords, and content (with HTML and JavaScript stripped).

Upload an image or document; AI extracts plain text directly. No login required.

Image generation powered by OpenAI's gpt-image-2 (GPT Image 2), upgraded with sharper detail and improved prompt adherence over gpt-image-1; also supports image editing with continuous refinement.

Nano Banana Pro image model.

Nano Banana 3.1 Flash, the lightweight tier.

Upload any image and the model reverse-engineers the prompt that would recreate it.

Google Gemini Omni Flash, a natively multimodal video generation model: create short videos from text prompts, or upload one PNG/JPG image as a reference (image-to-video).

Google Veo 3.1 text-to-video, a premium cinematic video generator; you can also upload an image as the first-frame reference (image-to-video).

SUNO music generator.

Upload a person photo plus a clothing image (or text description) and the model dresses the subject in the target outfit while keeping pose, face, and background intact.

Local ComfyUI image generation.

BytePlus Dola Seedream 5.0 Pro text-to-image, a flagship image generator with precise editing control, excellent Chinese prompt understanding, and strong photo-realism.

BytePlus Seedance 2.0 text-to-video.

A themed video generator that turns a name and a short cheer message into a personalized motivational clip.

One-click Christmas greeting video generator.

Chinese New Year greeting video generator.

BytePlus seed-tts text-to-speech.

Upload a 10-15 second sample of a person's voice and the model learns the timbre, then reads any text you provide in that cloned voice.

Translates text in any supported language into Traditional Chinese suitable for Taiwan/HK readers.

Translates text (especially Chinese) into natural American English.

English grammar and style checker.

Practice conversation with Sophie, a friendly female AI tutor. Replies are short text answers plus synchronized voice narration.

Practice conversation with Alex, a friendly male AI tutor. Replies are short text answers plus synchronized voice narration.

Build a private knowledge base; AI conversations cite your internal docs automatically.

Build your own chatbot; configure role prompts and knowledge bases for domain-specific AI.

Health-education assistant for generating patient-friendly explanations of medications, procedures, and care instructions.

Scans your text for simplified-Chinese characters and mainland-specific phrasing, highlighting each finding.

A quick Simplified-to-Traditional Chinese converter.

A quick Traditional-to-Simplified Chinese converter.