Kimi K3, Moonshot AI's 2.8-trillion-parameter flagship model, is now available on the Telnyx Inference API. It is the world's first open-source model in the 3-trillion-parameter class, built on Kimi Delta Attention and Attention Residuals with a 1M-token context window and native vision capabilities.

The model (ID: moonshotai/Kimi-K3) joins existing options like Kimi K2.6, GLM-5.2-FP8, and Mini Max M3. Key features include multimodal reasoning without a separate vision adapter, configurable reasoning effort (low, high, max), tool calling with structured output, and automatic prompt caching.

Telnyx emphasizes that competitive advantage is shifting to infrastructure, and K3 proves open-source models can rival closed-source frontier models from Anthropic and OpenAI on coding, reasoning, and agentic benchmarks. K3 runs on Telnyx-owned GPU infrastructure via an OpenAI-compatible API.

Pricing: cached input $0.27/1M tokens, input $2.70/1M, output $13.50/1M. Learn more in the Inference documentation or try it in Mission Control.