Groq

Removed Llama 4 Scout (17Bx16E) and Qwen3 32B from the main LLM pricing table, and Qwen3-VL 32B from the Enterprise-only table; Minimax M2.5 updated to Minimax M2.7 in the Enterprise-only table.

Kimi K2-0905 1T 256k removed from the main on-demand LLM pricing table (still priced identically in the separate Prompt Caching table) New Enterprise-only LLM tier introduced with Minimax M2.5 and Qwen3-VL 32B models, priced as contact sales (no public per-token rate)

PlayAI Dialog v1.0 TTS model ($50.00/M chars, 140 chars/s) replaced by Canopy Labs Orpheus English TTS model ($22.00/M chars, 100 chars/s) in the Text-to-Speech section.

GPT OSS 20B input token price decreased from $0.10 to $0.075 per million tokens (25% decrease) GPT OSS 120B output token price decreased from $0.75 to $0.60 per million tokens (20% decrease)

Prompt Caching feature added with 50% discount on cached input tokens for select models (Kimi K2, GPT OSS 20B, GPT OSS 120B) Mistral Saba 24B 32k model removed (previously $0.79 input/$0.79 output per million tokens)

DeepSeek R1 Distill Llama 70B speed increased from 275 to 400 tokens/sec 50% batch processing discount promotion ended (was doubled from 25% through end of April 2025)

Mixtral 8x7B Instruct 32k model removed (was $0.24/M tokens for both input and output at 575 tokens/second) DeepSeek R1 Distill Qwen 32B 128k model added with input price $0.69/M tokens and output price $0.69/M tokens at 140 tokens/second