Thinking Machines Releases Inkling 2 Small, a 594B MoE With 17B Active

- Thinking Machines Lab dropped Inkling 2 Small, a 594B MoE with 17B active parameters. It’s not a crypto token, but it’s efficient enough to make your GPU tears dry up. Scoring 31 on the Artificial Analysis Index, it crushes SWE-Bench Pro at 86.3%, proving agentic coding isn’t just vaporware. Tinker hosts it now; Apache 2.0 weights follow safety checks. DeepSeek and GLM are still playing catch-up in their price class. This is real infrastructure, not another rug-pull meme coin. While moonboys chase 100x leverage on nothing, actual compute power is being deployed. Stop waiting for Vitalik to fix your portfolio with a face twitch. Build with tools that work. Inkling 2 Small is live, cheap, and open-source soon. That’s the only alpha that matters.