sgcWebSockets .NET 2026.11.0 Latest
- 2026.11.0: 2026 November
- FixedAI: the default models now use current model ids, because the old ones are retired: the OpenAI chat bot transcription whisper-1 to gpt-transcribe and chat completion gpt-4o-mini to gpt-6-luna, the voice agent transcription gpt-4o-mini-transcribe to gpt-transcribe, the OpenAI speech request tts-1 to gpt-4o-mini-tts, the HTML AI chat component gpt-4o to gpt-6-astra and the default model of TsgcHTTP_API_OpenAI.Chat gpt-4o-mini to gpt-6-luna. The translation endpoint keeps whisper-1.
- FixedAI: the demos, help and samples use current model ids of OpenAI, Anthropic, Google Gemini, xAI Grok, Mistral and DeepSeek.
- NewAI: new TsgcAIVoiceTranslator, realtime speech to speech translation with OpenAI gpt-realtime-translate: streams the microphone, plays the translated speech and raises live transcripts of the source and the translated text; Language can change during a session and SendAudio accepts your own audio.
- FixedAI: TsgcAIOpenAITranslator no longer depends on whisper-1, which OpenAI removes with the /v1/audio/translations endpoint on 2027-02-26. It now transcribes with gpt-transcribe and translates the text with a chat model: new ChatModel (gpt-6-luna) and Language (English by default, so the output is the same as before).