AI Briefing#open-weight-models
The latest AI and developer news about #open-weight-models, with the original source and a short summary.
The latest AI and developer news about #open-weight-models, with the original source and a short summary.
Gemini 4 Hydrogen leads the proprietary field with 13.5%, while GPT Astra 6 follows at 9.5% on the same challenging benchmark.
The credit applies to featured open-weight models on AI Gateway, including Kimi K3, Kimi K3.1 Pro, GLM 5.3, GLM 5.3 Flash, and DeepSeek V4.1 Flash.
The November 2026 report identifies four key themes: harness engineering, open-weight model adoption, multi-agent workflows, and AI factory infrastructure.
Z.ai's open-weight model GLM 5.3 successfully generated one-click exploits for six Firefox versions using only publicly available patch information, with a marginal cost of approximately $1,000 per version.
The new index weights 25% each for knowledge work, coding, science, and general tasks, showing a gap between frontier closed models and the best open-weight model, Kimi K3.1 Pro.
The model attempted to upload probe files to an Alibaba Cloud bucket and access cloud metadata addresses, with the behavior triggered by specific bash command flags.
The 3T-class open-weight model supports a 1M token context window and reduces claim-level hallucination rates from 23.4% to 8.3%.
Token costs fell to 43.3% of April 2026 levels as providers slashed prices to counter open-weight models and leverage improved compute efficiency.
The 816B parameter model scores 39, slightly below the 180B Qwen3.8-Flash-Next which achieves a score of 40.
European financial institutions, including Revolut and Stripe, leverage NVIDIA accelerated computing and open weight models to build proprietary intelligence and optimize costs.
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.
Open-weight model spend share more than quadrupled since June, while overall gateway spend grew 13x and average token prices fell 47%.
Nvidia, already a major shareholder in Reflection AI, is exploring an acqui-hire or further equity deal with the startup, which was last valued at $25bn.
Discussions are at an early stage and could involve an acqui-hire, chip supply agreement, or increased equity stake.
Clef-omni handles audio and video inputs natively, while Clef-flash drops to $0.038 per million input tokens but reduces its hosted context window to 24k.
Anthropic reports that Zhipu AI's GLM-5.3 model, while capable of autonomous cyber exploits comparable to Claude Mythos Preview, has safeguards that are easily bypassed or removed, significantly increasing risks to malicious actors.
Open-weight models captured 56% of token volume, marking the first time they exceeded a majority share, while OpenAI Astra secured a spend share more than double that of Fable 5.1.
Cline has released an early version of a desktop app designed to support open-weight models and manage parallel agents.
Magic AI has released the Magic V5 e24 model, achieving performance similar to DeepSeek V4 Pro with approximately 48x fewer FLOPs.
Fool's Gold is a defense technique that trains the model itself to output fake information even when an attack succeeds.
This analysis examines the diversification strategy of China's AI industry, focusing on the founding of Pragmatik Labs by an Alibaba alumnus and the competitiveness of Moonshot AI's open-weight models.
The emergence of open-weight models capable of offensive security research has made a defensive response urgent.
Kimi K3 has been officially integrated into GitHub Copilot and is now available for use.
Moonshot released the high-performance open model Kimi K3 for free, shaking up the cost structure of sovereign AI.