Kimi K3 is a massive new open AI model with 2.8 trillion parameters, native vision and a one-million-token context window. Its creator claims it can compete near the frontier of coding, reasoning and knowledge work - but its potential goes far beyond benchmark scores. In this Web News, Matt and Mike discuss what happens when companies can operate powerful AI without depending entirely on OpenAI, Anthropic or another hosted provider. Could open models lower costs and unlock better AI products, or will their customizable guardrails create new safety and regulatory concerns?
Editor’s note: This episode was recorded shortly after Kimi K3 was announced. Kimi describes K3 as an open model, with the full model weights scheduled for release by July 27, 2026. Kimi recommends 64 or more accelerators for full deployment, so the complete model is not designed to run on a typical gaming (or high-end) PC. Benchmark results discussed in this episode are primarily based on Kimi’s initial evaluations and have not yet been fully independently verified.
Kimi has introduced Kimi K3, a massive open AI model designed to compete near the frontier of coding, reasoning and knowledge work. With 2.8 trillion parameters, native vision and a one-million-token context window, it represents another major step forward for open models.
But its benchmark performance may not be the most important part of the story. If corporations can host and control increasingly powerful models themselves, they could reduce their dependence on companies like OpenAI and Anthropic. That could lower costs and enable new AI products - but it could also make safeguards easier to modify and create difficult questions about security and regulation.
