Home/Events/PrismML Releases Bonsai 2 27B Model with Advanced Compression Technology

PrismML Releases Bonsai 2 27B Model with Advanced Compression Technology

Developing

Confidence

80%

Impact: 70%

Updated 10h ago

Consensus Brief

PrismML has launched its latest model, Bonsai 2 27B, which compresses the Qwen3.8 27B model down to 5.9 GB, allowing it to run on PCs and potentially high-end smartphones. The new model achieves 98% of the benchmark performance of its larger counterpart, showing significant improvement in compression technology. The startup aims to apply this technique to even larger models in the future.

Sourced from

TechCrunchT2

Primary: TechCrunch

What Changed Since Last Update

10h ago

The Bonsai 2 model improves benchmark performance from 95% to 98% compared to the original Qwen model.

Claim Ledger

5 claims tracked across sources

Confirmed Fact

PrismML raised a $22.25 million seed round.

Confirmed Fact

Bonsai 2 matches 98% of Qwen’s aggregate benchmark scores.

Confirmed Fact

The original Bonsai model has been downloaded over 11 million times.

Confirmed Fact

PrismML's smaller models have been downloaded another 2.6 million times.

Confirmed Fact

PrismML's compression technique uses ternary weights.

Role-Based Impact Analysis

Source Timeline

1 source corroborating

PrismML hopes its tiny LLM could change how we all use AI

TechCrunch·10h ago