EmpirioLabs AI

EmpirioLabs AI published a blog post on September 23, 2026 documenting three new Qwen Audio 3.1 speech-to-text endpoints available on its platform: qwen-audio-3-1-asr for batch transcription of recorded audio via POST /v1/audio/transcriptions, qwen-audio-3-1-asr-stream for live captions with partial and completed trans All three endpoints are described as available both through the EmpirioLabs API and in the Playground, giving developers a single path to test batch and live transcription. The post also clarifies event semantics: conversation.item.input_audio_transcription.delta carries the in-progress sentence for ASR Stream only, wh

EmpirioLabs AI published a blog post on August 26, 2026 announcing that it is now a listed provider on Opper, a Stockholm-based European AI gateway. Through the listing, Opper users can call EmpirioLabs' Native Inference text models as direct endpoints on the Opper platform, with EmpirioLabs' provider page live at oppe The blog enumerates the Native Inference text models EmpirioLabs exposes through Opper: Qwen3.5 4B, Qwen3.5 9B, Qwen3.6 35B A3B, Qwen3.8 27B, Gemma 4 26B-A4B, and Muse Glimmer 30B, all served on infrastructure EmpirioLabs operates. It positions the partnership as a European entry point for frontier and open-weights mod

EmpirioLabs AI's model page for Qwen3.8 Omni Flash Realtime, released September 21, 2026, attributes the model to Alibaba Cloud and positions it as a real-time multimodal speech-to-speech endpoint hosted on EmpirioLabs' WebSocket /v1/realtime interface. The page specifies a 192K-token context window, a 65,536-token max Configuration is handled through session.update events over the WebSocket rather than request parameters, and the page notes that audio travels as base64 16-bit PCM in both directions. Because browsers cannot set Authorization headers on a WebSocket handshake, the documentation recommends opening the connection from a

EmpirioLabs AI published a documentation changelog on July 7, 2026 announcing two new platform features for developers. The new Batch API lets users submit large asynchronous jobs at a 35% discount off list price: developers upload a JSONL file of requests to /v1/files, create a batch via /v1/batches, and poll for the The same July 7, 2026 release also introduced a Gemini-compatible endpoint at https://api.empiriolabs.ai, allowing the google-genai SDK (Python or JavaScript) and any Gemini-native client to call EmpirioLabs chat models through generateContent and streamGenerateContent with an EmpirioLabs API key. All chat parameters c