Kimi K2.7 and K3, GLM 5.2 and Qwen3.8 Max, free for everyone for three days
Four open-weight flagships in the model menu — and through August 8 they cost nothing, on every account. Clusy ❤️ open source.
Kimi K2.7 and K3, from Moonshot, GLM 5.2, from Z.ai, and Qwen3.8 Max, from Alibaba, are all in the model menu today. All four are open-weight: the labs that trained them published the weights for anyone to run, inspect, and build on. For three days, all four are free on Clusy — every account, including free ones, and turns on them don't draw down your usage allowance.
That fits what a Clusy session is. When the agent plans an experiment, writes the cells, runs them, and reads the results, the transcript grows fast — data schemas, training logs, tracebacks, the three branches you tried before the one that worked. K2.7, K3, GLM 5.2 and Qwen3.8 Max are strong, current models that can hold that context and reason through it, and for the next three days the cost of handing them a real problem is nothing.
Free for three days
Open the model selector, pick Kimi K2.7, K3, GLM 5.2 or Qwen3.8 Max, and give it something real: a fine-tune, a messy dataset, a paper you want reproduced. No subscription, no card — a free account is enough.
The free window ends on August 8 at 00:00 UTC.
Clusy ❤️ open source
Open weights are why we can do this at all. The three labs did the expensive part — trained the models and released them — and for three days we're taking the paywall off work that was published openly. We've given some of it back too. Making the open models free to use, even for a few days, is the same instinct pointed the other way.
If you build something with them before the window closes, we'd like to see it.