What field is model compression part of?

Model compression is part of Efficient inference and compression.

Is model compression ready for products?

Its papers score 2.6 of 5 for readiness for products, read from what the papers themselves describe.

Which companies work on model compression?

Researchers at Meta publish on it.

Model compression

Part of Efficient inference and compression

Model compression is losing ground in AI research: 0.3% to 0.2% of new AI papers, -0.0 points. It ranks #172 of 214 by gain in share. As of Oct 10, 2026.

How we measure

Change in share: -0.0 pts
Share, last 3 months: 0.2%
Papers, last 3 months: 76
Ready for products: 2.6 / 5

More in Efficient inference and compression

Kv cache reuse0.1% → 0.2%+0.1 pts Kv cache compression0.2% → 0.3%+0.1 pts Load balancing0.1% → 0.1%+0.1 pts Visual token compression0.1% → 0.1%+0.1 pts Neural network quantization0.0% → 0.1%+0.0 pts Diffusion model acceleration0.2% → 0.2%+0.0 pts Structured pruning0.1% → 0.1%+0.0 pts Mixed precision quantization0.1% → 0.1%+0.0 pts Visual token pruning0.2% → 0.2%+0.0 pts Adaptive computation0.2% → 0.2%+0.0 pts

Companies working on it

1Meta1 papers

Model compression: quick answers

No. Its share of new AI papers went from 0.3% to 0.2%, with 76 papers in the last 3 months.