AI Briefing#model-release
The latest AI and developer news about #model-release, with the original source and a short summary.
The latest AI and developer news about #model-release, with the original source and a short summary.
The new 8B and 70B models add native image and audio understanding, thinking mode, and a 4x larger context window under an Apache 2.0 license.
The new flagship model features a 600B/27B architecture and supports a 1M token context window.
The model features 42B active parameters per token, a 1M-token context window, and supports text, image, audio, and video inputs.
The update introduces opt-in 'danger zone' flags for image support, a new 'box' type, and extended context lengths, alongside a new model version focused on multilingual improvements.
The model scores 59.3 on the internal PI Knowledge Work benchmark at $0.13 per task, undercutting Claude Fable 5 and GPT-5.6 Sol while maintaining competitive quality.
The 74B parameter MoE model is packed into 24GB using 2-bit quantization, with a 4GB variant releasing Oct. 23.
The 3.66B model achieves a 55.4 logic composite score, effectively tying with Qwen3-8B and offering high-speed CPU inference.
The new Muse Wonder series introduces Active Memory and Active Focus features, available now via the Meta Model API and Muse Code.
The new model achieves 85.9% of the BF16 baseline score while being 13.9× smaller in file size.
The new model achieves a Decision Index score of 61.56, surpassing Jev by more than 3.5 points using the same Qwen3.8-27B backbone.
The model generates recognizable 64x64 images without using oversized text encoders or VAEs.
The new model, available in Microsoft Foundry, delivers calibrated probability scores for routing and classification tasks at 35 times the speed of GPT-6 Sol.
The new model costs around 75% less than Haiku 4.5 and introduces adjustable effort settings for the first time in the Haiku class.
Reflection AI has introduced Beam, a 501B-parameter open-weight MoE model trained on 10.5K NVIDIA GB300 GPUs, claiming high inference efficiency compared to larger models like GLM 5.2 and Kimi K3.
The 9B multimodal model claims 8 of 9 top scores in its size band on spatial benchmarks and includes an open-sourced data pipeline.
The 8B model, based on Qwen3-8B, generates reports in an average of 51.1 seconds compared to 178.5 seconds for the previous Claude-based Thinking mode.
RelateAnything supports an inference speed of 17.2ms per image on the RTX 3080 Laptop GPU using TensorRT FP32 and has released the RA-4M training corpus, which includes over 474,000 images and 4.3 million relationships.
Victoria achieves 70.0% on Terminal-Bench 2.1 using REAP pruning, while Maple improves Canadian regulatory accuracy from 6.6% to 21.8%.
GPT-6.1 Sol is available via API and ChatGPT Work/Codex, offering cached input pricing at $0.10 per million tokens and matching GPT-6 Astra on agentic coding tasks at significantly lower cost.
The model features 320B total parameters with only 15B activated per token, optimized for reasoning and multi-step tool use.
The release includes DFlash draft models for the 9B, 35B-A3B, and 397B variants of Ornith-1.5.
The 550B-parameter model scored 535.4 out of 600 on the IOI 2026 problem set, surpassing the top human contestant's score of 498.27.
The new model uses Glow-TTS architecture and targets edge devices and CPU inference.
The new models accept shared state, named questions, and optional images to return answer distributions in a single forward pass.