ninjachat api · status

A quiet hour.

Live health of every model on the v1 API, measured from real request traffic. Updated every minute.

as of Fri, 21 Aug 2026 13:15:00 UTC

Google models

Quiet

gemini-3-flashQuiet—

gemini-3.1-proQuiet—

gemma-4-31bQuiet—

Tencent models

Quiet

hy3Quiet—

Meta models

Quiet

llama-4-maverickQuiet—

Anthropic models

Quiet

claude-sonnet-4.6Quiet—

claude-sonnet-5Quiet—

ByteDance Seed models

Quiet

seed-2.0-proQuiet—

seed-2.0-codeQuiet—

seed-2.0-miniQuiet—

Qwen models

Quiet

qwen-3.6-35b-a3bQuiet—

qwen3-next-80b-a3bQuiet—

qwen-3.8-maxQuiet—

qwen-3.8-27bQuiet—

NinjaChat models

Quiet

uncensored-aiQuiet—

Xiaomi models

Quiet

mimo-v2.5Quiet—

OpenAI models

Quiet

gpt-oss-20bQuiet—

Mistral AI models

Quiet

mistral-small-4Quiet—

Image generation

Quiet

seedreamQuiet—

Video generation

Quiet

veo-3.1-fastQuiet—

Methodology

Measured from live API traffic over the last hour; we publish what we measure — no synthetic uptime claims. Failures count HTTP 5xx only: client errors and rate limits (4xx, 429) are excluded, and models with fewer than 10 recent requests are shown as “quiet” rather than claimed operational.

Programmatic access: GET /api/v1/health · Live benchmarks