Alibaba's Qwen team released Qwen3.8-Omni-Flash on 2026-09-18 — its first omni-modal model built around agentic capabilities. It natively accepts text, image, audio and video input and returns text-only output (no speech synthesis), with a 1M-token context window (max output 131,072 tokens). Served as a proprietary hosted model on Alibaba Cloud (Model Studio / DashScope, OpenAI-compatible) at $0.15/1M input and $0.47/1M output. It targets agentic audio-video understanding (long-video segment location, meeting summaries, video research, vlog auto-edit, short-video translation, movie recaps) with function calling, web search, structured outputs and caching. Alibaba reports >26% average gain across 30 evals vs Qwen3.5-Omni-Plus, ~51.8% fewer tokens on OmniVideoBench in agent-perception mode, and ~89% lower video input cost; figures are self-reported. Ships with open-source Qwen-MM-Plugins.
PrismML released Bonsai 2 27B on 2026-09-17 under Apache 2.0 — a ternary (1.58-bit) compression of Alibaba's Qwen3.8 27B that shrinks the 27.8B-parameter multimodal model to 5.9 GB (a 9-10x memory reduction) so it runs locally on PCs and high-end phones. It reports an 83.9 aggregate across a 20-benchmark suite (reasoning, math, coding, instruction following, vision, agentic tool use), retaining 98.2% of Qwen3.8 27B's 85.4, up from the first Bonsai 27B's 95% in July, and reaches ~143 tok/s on an RTX 5090. Free weights on Hugging Face. First tracked model from PrismML (Pasadena, CA; Caltech spinout led by Babak Hassibi). Self-reported figures.
TypeSafe AI released Jev in early access — the first model in its "System One" line, built for automation workflows rather than chat. Jev maps unstructured state to typed probabilistic decisions and emits parallel structured outputs (JSON / tool calls), trained with Reinforcement Learning for Calibrated Decisions (RLCD), and is positioned as a "frontier-intelligence function call" for agents, classification, and tool use. Proprietary (conditional commercial use); weights not released; context window undisclosed. Priced at $0.042 /1M input with output free on a single TypeSafe AI serverless route at launch. Vendor figures unverified independently.
At Dreamforce, Salesforce and NVIDIA announced Koa, Salesforce's first CRM reasoning model for Agentforce, built by post-training NVIDIA Nemotron 3 Super on a proprietary synthetic dataset modeled on ~27 years of CRM deployments (no customer data used). Koa reasons through multi-step enterprise workflows and uses tools to act; on Salesforce's CRM benchmark it matches or exceeds leading models on CRM actions with roughly 3x fewer errors. Salesforce controls the weights and runs inference inside its own trust boundary (weights not released). Available to select pilot customers at launch; general availability expected winter 2026 in U.S. regions.
Agnes AI, a Singapore omni-modal foundation-model lab, surfaced Agnes 3.0 Flash in mid-September 2026. The disclosed open-weights preview checkpoint (Apache 2.0) is a 33B hybrid-attention model with a 262K context and text/image/video input: 54 of 72 decoder layers run a gated delta rule while 18 use standard grouped-query attention, holding down the KV cache at long context. Runs at bf16 on a single H100/H200-class GPU. The production Agnes 3.0 Flash served via the company's API is a separate checkpoint with a 1M-token context. First model tracked from this org. Vendor figures unverified at launch.
From 04:00 UTC on 2026-09-14 DeepSeek routes every `deepseek-v4-pro` request to V4.1-Flash, billed at V4.1-Flash rates, and says this will continue until V4.1-Pro launches. The id still answers, making this a redirect and a deprecation rather than a retirement — DeepSeek reports V4.1-Flash beats V4-Pro on performance, cost, speed, and total time.
Inference.net listed Schematron V2 Turbo and Small, a pair of 3B HTML-to-JSON extraction models in its "workhorse model" line — small purpose-built LLMs sold on cost per unit of work. Both are schema-driven (the extraction target goes in a JSON schema via response_format, not the prompt) with 128K context; Turbo is throughput-optimized at ~4.14 req/s on one H100 and $0.03/$0.15 per Mtok, Small trades throughput for quality on complex schemas at $0.05/$0.23. Proprietary and API-only via Inference.net and OpenRouter. First models tracked from this org.
Shanghai AI Laboratory released Atria Dawn Preview weights-first on 2026-09-11 — code and checkpoint appeared on GitHub / Hugging Face with no announcement, followed ~three days later by a 140-author technical report. It is a 744B-parameter agentic Mixture-of-Experts model built on GLM-5.2, aimed at long-horizon research agents that take a method from the literature to executable experiments, reproducible metrics, and an inspectable report. MIT license, open weights, 256K context. On the lab's own 16-benchmark table it leads on five tasks incl. AutomationBench 53.8, BrowseComp 92.5, DeepSearchQA 96.0, BFCL v4 77.0 and CyberGym 86.5. Self-reported figures.