TinyLeague updates benchmarks: drops AA Index, adds KV efficiency metrics
TinyLeague, a tracker for small open-weight AI models, updated its benchmarking platform on October 28, 2026. The latest changes include removing the AA Intelligence Index and detail tables, while adding KV cache efficiency metrics and stricter cutoff criteria. The platform now focuses on KV cache requirements (measured at 128K tokens) and rates models on a 0–100 log scale for efficiency,…
Key points
- Removed AA Intelligence Index and detail tables; added KV cache efficiency metrics (0–100 scale)
- Two models unrated due to missing active-parameter data; heatmap hides low-entry columns under filters
- Added 10+ new models, including LiquidAI/LFM3-1.7B and MiniCPM-V-4.7-35B (first open-source on-device MoE VLM)
Key additions include LiquidAI/LFM3-1.7B, Sage-A1.5, and XHToken/Spark-X2.6-4B, alongside vision-capable models like Ornith-1.5-122B and MiniCPM-V-4.7-35B, the latter noted as the first open-source on-device MoE VLM. Older entries like Nemotron-3-Nano-Omni-30B-A3B-Reasoning were removed. The updates aim to improve clarity and comparability for researchers evaluating small models.
Model pages: LFM3-1.7B → · Sage-A1.5 → · Spark-X2.6-4B →
Important update to TinyLeague (AA out, KV efficiency in)
tinyleague.nunodonato.com · 28 October 2026Loading the full article…
This text was published by tinyleague.nunodonato.com. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Reddit discussionreddit.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- OpenAI and Anthropic claim their AI models refuted the Exponential Time Hypothesis · 13 src
- Cprisp framework predicts circRNA-RBP interactions for neurodegeneration research · 1 src
- Analysts followed AI agents' leans 86% of the time in study · 1 src
- Artificial Analysis releases Intelligence Index v5 with private datasets · 1 src
- DeepSeek’s R1 model with guideline knowledge base improves junior doctors’ arthroplasty decisions · 1 src
Comments
via GitHub Discussions