Model status

25 of 25 models operational

Model	Status	Latency	Fallback
deepseek DeepSeek v4-flash	operational	0.9s	none needed
google Gemini Flash-Lite	operational	0.9s	none needed
zhipu GLM-Flash	operational	0.9s	none needed
alibaba Qwen flash	operational	0.9s	none needed
together Llama 4 Scout	operational	0.9s	none needed
yandex YandexGPT Lite	operational	0.9s	none needed
yandex Alice Flash	operational	0.9s	none needed
anthropic Claude Haiku 4.5	operational	0.9s	none needed
openai GPT-5.4-mini	operational	0.9s	none needed
google Gemini 3.5 Flash	operational	0.9s	none needed
moonshot Kimi K2.5	operational	0.9s	none needed
xai Grok 4.3	operational	0.9s	none needed
mistral Mistral Medium 3.5	operational	0.9s	none needed
sber GigaChat Pro	operational	0.9s	none needed
yandex Alice AI LLM	operational	0.9s	none needed
yandex YandexGPT Pro	operational	0.9s	none needed
anthropic Claude Sonnet 5	operational	0.9s	none needed
google Gemini 3.1 Pro	operational	0.9s	none needed
openai GPT-5.6 terra	operational	0.9s	none needed
xai Grok 4.5	operational	0.9s	none needed
moonshot Kimi K3	operational	0.9s	none needed
zhipu GLM-5.2	operational	0.9s	none needed
sber GigaChat MAX	operational	0.9s	none needed
openai GPT-5.6 sol	operational	0.9s	none needed
anthropic Claude Opus 4.8	operational	0.9s	none needed

Model calendar

deepseek-chat and deepseek-reasoner model names retire 2026-07-24

Moonshot V1 line shuts down 2026-08-31

Claude Sonnet 5 API price rises to $3/$15 per M 2026-09-01

Jul 20, 12 min

GPT Image queue delays, image jobs held and completed, no tokens lost.

Jul 20, 41 min

Kimi K3 error rate spiked, traffic auto-routed to GLM-5.2, users notified in-thread. Resolved by provider.

Coefficient changes are announced 14 days ahead, here and in the app.