HHG for TranslatePress

HHG for TranslatePress extends TranslatePress with best-in-class AI translation engines. Pick from four engine families instead of a single provider.

Supported Translation Engines:

Google Gemini AI

Powered by Google’s latest Gemini models. Choose from Flash-Lite (fastest & cheapest), Flash (balanced), Pro (highest quality), or 3 Flash Preview (cutting edge). API key from Google AI Studio.

OpenAI Compatible Platform

A universal engine supporting seven OpenAI-compatible API providers through a single, unified interface:

• OpenAI — GPT-4o Mini, GPT-4o, GPT-4.1, o4-mini, GPT-4 Turbo
• Azure OpenAI — Bring your own deployment endpoint
• Groq — Ultra-fast inference with generous free tier (Llama 4, Mixtral, Gemma)
• Together AI — Open-source model hosting (Llama 4, DeepSeek V3, Qwen3)
• OpenRouter — Multi-provider routing (access OpenAI, Anthropic, Google models)
• Ollama — Run locally, zero API cost (Llama 3.2, Qwen3, Mistral, Gemma 3)
• Custom — Any OpenAI-compatible endpoint you provide

Each platform preset includes curated model suggestions, platform-specific descriptions, and automatic API configuration. The model input supports both free-form text entry and one-click suggestion chips — type any model name or click to select.

DeepSeek AI

DeepSeek V4 Flash for fast, affordable translations and V4 Pro for professional-grade quality. Optimized with thinking disabled for faster throughput and lower cost.

ZhiPu AI GLM

Six translation strategies: General, Paraphrase, Two Step, Three Step, Reflection, and Chain of Thought. Choose speed or depth depending on your content.

Key Features:

• Industry / Domain Description — Describe your site’s context once (industry, tone, terminology rules) and it’s injected into every translation request for consistent, accurate output across all engines
• Three-layer context system — Sliding-window surrounding strings, page-level key term extraction, and industry prompt for terminology consistency
• Parallel async translation — All chunks dispatched simultaneously via cURL multi-handle for maximum speed
• Two-round retry logic — Failed strings automatically retried in a second pass with lower temperature for maximum completeness
• Intelligent caching — 30-minute transient cache per chunk for instant repeat translations
• Adaptive chunk sizing — Chunk size auto-adjusted based on average string length per page
• Platform-aware configuration — Timeout, temperature, and chunk size tuned per model family via heuristic detection
• Native TranslatePress integration — Settings panel styled to blend seamlessly with TranslatePress’s own UI, with smooth CSS transitions
• Code security — All inputs sanitized, outputs escaped, nonces verified, SSL enforced, following WordPress best practices

External Services

This plugin connects to external AI translation services to provide automatic translation functionality. The following third-party services are used:

Read the full description on the official page →