Benchmarks - Page 13
Stories from Benchmarks come first, newest first, then stories from connected topics (AI Models, LLM, AI, coding agents, preprint, open weights), each labelled with where it comes from. This view is not indexed by search engines.
Latest
Salesforce Says AutoR&D Agents Improved Its Embeddings
Salesforce says SFR-AutoR&D agents raised nDCG@10 by 6.13 points in research and 4.69 in a production-supported run. The model is not yet customer-facing.
Runway Opens Praxis-1 Weights Under the Apache 2.0 Licence
Runway has released Praxis-1, a 27-billion-parameter world action model, with open weights and fine-tune code under Apache 2.0 for commercial use.
Gemini Decisions API Launches With Gemini Decision 1
Google launched the Gemini Decisions API and Gemini Decision 1, at $0.10 per 1M input tokens with no output tokens billed, on 11 November 2026.
Baidu Releases ERNIE 5.5 Flash Preview on the Qianfan API
Baidu has released ERNIE 5.5 Flash Preview, a 280B-total, 14B-active model with a 1M-token context, on Qianfan. Apache 2.0 weights are planned.
Google Cloud Adds Meta Muse Ember, Spark 1.3 and Wonder
Meta’s Muse Ember, Spark 1.3 and Wonder are now on Google Cloud as managed APIs, under separate Meta terms, not as a Google product.
SpeakLeash Releases Bielik 11B v3.1 in 56 Languages
SpeakLeash and ACK Cyfronet AGH have released Bielik 11B v3.1, a 56-language open model from Poland. Bielik-One, due 12 November, is not out yet.
IFM Releases PAN 2 Open World Model With Training Code
IFM has released PAN 2, an open world model that turns a scene and a text action into 20-second videos. Its benchmark rankings are IFM’s own.
MiniMax H3-Image Released; Licence Excludes the UK
MiniMax has released MiniMax H3-Image open weights, but the community licence excludes the UK, EU, US and Republic of Korea unless separately granted.
Australia Plans an AI Regulator and Frontier Safety Laws
Australia plans an AI regulator and frontier safety laws. The legislation and regulator are proposals, with no bill text, timetable or penalty amounts yet.
LG AI Research Releases Free API Access to K-EXAONE 2.5
LG AI Research has opened a free API for K-EXAONE 2.5. Internal agentic scores are company-reported and trail GLM-5.3 on two of four tests.
Latest on TSN
Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.
