Groq adds text-to-speech, browser automation, and new models
Groq launched Orpheus text-to-speech billed per 1M characters ($22 English / $40 Arabic), a Browser Automation tool at $0.08/hour, and new models including GPT OSS Safeguard 20B, Qwen 3.6 27B, and Kimi K2.
Per-token LLM inference, Whisper transcription, and built-in tools (web search, visits, code execution). No speech synthesis; core tools capped at code execution.
Adds a text-to-speech SKU (Orpheus V1 English $22.00/1M chars, Orpheus Arabic Saudi $40.00/1M chars), Browser Automation tool at $0.08/hour, new serverless models (GPT OSS Safeguard 20B $0.075/$0.30, Qwen 3.6 27B $0.60/$3.00, Kimi K2 $1.00/$3.00 with $0.50 cached input), and enterprise-only Minimax M2.5 and Qwen3-VL 32B. Core per-token rates unchanged.
Groq expanded GroqCloud beyond text and transcription into speech synthesis. The new Canopy Labs Orpheus text-to-speech models introduce a per-1M-character billing dimension ($22.00 for V1 English, $40.00 for Arabic Saudi), giving Groq a full audio round-trip on LPU silicon — Whisper ASR in, Orpheus TTS out.
On the agentic side, a new Browser Automation built-in tool joins the per-use rate card at $0.08/hour, alongside the existing web search ($5–$8 per 1,000 requests), website visits ($1 per 1,000 requests), and code execution ($0.18/hour). The model catalog also grew: GPT OSS Safeguard 20B ($0.075/$0.30), Qwen 3.6 27B ($0.60/$3.00), and Moonshot Kimi K2 ($1.00/$3.00, with cached input at $0.50) are now published, plus enterprise-only Minimax M2.5 and Qwen3-VL 32B behind contact-sales.
Core per-token rates for Llama 3.1 8B, GPT OSS 20B/120B, Qwen3 32B, and Llama 3.3 70B held stable — this is a packaging and catalog expansion, not a repricing.
Groq added a text-to-speech product line billed per 1M characters (Canopy Labs Orpheus V1 English at $22.00/1M chars, Orpheus Arabic Saudi at $40.00/1M chars), a new Browser Automation built-in tool at $0.08/hour, and several new models (GPT OSS Safeguard 20B $0.075/$0.30, Qwen 3.6 27B $0.60/$3.00, Kimi K2 $1.00/$3.00 with $0.50 cached input) plus enterprise-only Minimax M2.5 and Qwen3-VL 32B. Core per-token rates for Llama, GPT OSS, and Qwen3 held stable.