PrismML Fans Want Bonsai 32B

This is less a product release than a demand signal: the community likes the Bonsai idea enough to immediately ask for a much larger tier. The catch is that Bonsai’s whole pitch is efficiency, so a 32B variant would have to prove it can preserve that edge instead of just becoming another big local model.

• –PrismML’s official messaging currently centers on 1-bit Bonsai 8B, 4B, and 1.7B, not a 32B model
• –A 32B Bonsai would raise the bar on memory, throughput, and hardware compatibility, especially for local inference
• –The Reddit thread shows strong appetite for a higher-capacity model, which is useful product feedback even without an announcement
• –If PrismML does ship a 32B, the key question will be whether the 1-bit approach still delivers enough quality-per-watt to matter
• –For now, this reads as community speculation and wishcasting, not a confirmed roadmap item