The developer unveiled Inkling 2 Small, a Mixture-of-Experts model with 594 billion total parameters and 17 billion active, supporting image inputs and variable effort levels. The model scored 31 on the Artificial Analysis Intelligence Index, a notable jump from the company's prior generation, with gains attributed to more efficient pre-training and a unified reinforcement learning run. It is currently accessible only through the Tinker platform with closed weights, ahead of a planned Apache 2.0 open-weight release next week pending safety testing.
thinkingmachines.ai
· 2026-11-06
Google introduced EmbeddingGemma 2, a 740M-parameter embedding model that maps text, code, images, video, and audio into a single unified embedding space. It supports flexible output dimensions from 768 to 128 via Matryoshka Representation Learning, offers an 8K context window—four times larger than the text-only EmbeddingGemma—and is released under the Apache 2.0 license.
twitter.com
· 2026-10-06
Google launched EmbeddingGemma 2, a follow-up to its original EmbeddingGemma model that reportedly surpassed 20 million downloads. The new 740-million-parameter model is built on the Gemma 4 architecture, released under an Apache 2.0 license, and extends beyond text to embed code, images, video, and audio in a single shared space for on-device inference.
blog.google
· 2026-10-06
Strands harness is a preassembled agent harness that developers can run locally or deploy to cloud providers with a single line of Python or TypeScript code, released under an Apache 2.0 license. It is designed as a general-purpose agent rather than a coding-specific one, and its makers report it costs 28% less than comparable harnesses across six benchmarks while maintaining similar or better accuracy, and 77% less than Claude Code on the Fable 5 benchmark while scoring higher on Terminal Bench 2.1.
strandsagents.com
· 2026-09-23
A new ternary-quantized model, Ternary Bonsai 2 27B, has been released, built on Qwen3.8 27B and using {-1,0,+1} weights with FP16 group scaling to shrink the model to about 1.76 effective bits per weight and a 5.9GB footprint. Despite being over 9x smaller than its full-precision counterpart, it retains 98.2% of aggregate benchmark performance across reasoning, coding, vision and agentic tasks, and supports a 262K-token context window under an Apache 2.0 license.
prismml.com
· 2026-09-17
GoKawiil