All AI news

AIModelScope released Corvus-Gov-3B, a compact model tuned for Chinese policy Q&A, public-service consultation, and internal government or enterprise assistants. It was fine-tuned on one million Chinese government-domain dialogue samples and built on Llama 3.2 3B Instruct using LoRA SFT via LLaMA Factory. The model is released under Apache 2.0.

AIOpenBMB reports that its MiniCPM5-2B model runs at 37 tokens per second on an iPhone Air. The post presents this as evidence that small open multimodal models can run on mobile devices without a cloud GPU, with NobodyWho noting the model is available in its Chat app.

AIUnderdog has released Saluki 27B under Apache 2.0, a 2-bit GGUF of Qwen3.8-27B that fits in 7.89 GB, versus 54 GB for the full BF16 model. On Underdog Bench, Saluki scores 88 against 84 for the full model, and it raises parallel tool-call accuracy to 42 from 35. It runs on stock llama.cpp, but math and reasoning drop sharply, with AIME 2025 at 79.2 versus 96.7.

AITencent has open-sourced Youtu-Parsing-Omni, a 5B-parameter omni-modal model that outputs a single structured JSON covering layout, text, tables, formulas, ASR, OCR, and video segments. It scores 96.96 Overall on OmniDocBench, the highest among the compared models, and ships with weights on Hugging Face, a vLLM plugin, and inference examples.

AINVIDIA published the Agile One S SSD Pick deployment model, GR00T N1.7 checkpoint 40000, on Hugging Face for SSD pickup tasks. The repository includes five ONNX graphs with external tensor files and two existing TensorRT BF16 engines, with original configurations and build metadata, but no retraining or re-export was performed. The files are not a robot deployment or safety qualification, and engine compatibility depends on the target GPU and TensorRT environment.

AINVIDIA has released a deployment model for Agile One S SSD pickup, based on GR00T N1.7 checkpoint 58000 and using three cameras: ego, left wrist, and right wrist. The repository republishes ONNX graphs, external tensor files, and two existing TensorRT BF16 engines without retraining or re-export, and the original export reported a numerical warning that full FP32, node, and BF16 parity did not pass all tolerances. The files are not a certified robot deployment or safety qualification.

AINVIDIA published the Agile One S Walk GR00T N2 checkpoint 1680, a walking deployment model with four cameras, on Hugging Face. The repository includes ONNX graphs, TensorRT BF16 plans/engines, and the original checkpoint files, republished without retraining or re-export. The shared Cosmos-Reason1-7B dependency and the Isaac/GR00T runtime must be set up separately, and the files are not a robot safety qualification.

AINVIDIA published the nvidia/agile_one_s_place_ssd_n17_24050 repository on Hugging Face, containing a GR00T N1.7 checkpoint 24050 model for placing an SSD with three cameras. The repository includes five ONNX graphs with external tensor files and two existing TensorRT BF16 engines, republished without retraining, re-export, or engine rebuild. Engine compatibility depends on the target GPU and TensorRT environment, and the files are not a robot deployment or safety qualification.

AIJetBrains released Mellum2.1, a 12B mixture-of-experts coding model with 2.5B active parameters under Apache 2.0, emphasizing agentic programming. Under high load, its inference throughput in tokens is nearly twice that of Qwen3.5-9B in JetBrains' comparison, and multi-token prediction (MTP) speeds single-request responses by about 1.6x. The model is available on Hugging Face for local or private-infrastructure deployment, with GGUF and vLLM MTP support announced for later.

AIOdyssey announced the Odyssey-3 series of foundation world models, with Odyssey-3 Pro scoring 66.1 on the Physics-IQ Verified video-to-video benchmark, the highest recorded on that leaderboard. The series includes a standard version balancing physical accuracy and generation cost, and a Pro version with stronger physics prediction. The preview supports first-person and third-person navigation and lets users move the camera, take actions, or trigger events while the model predicts environmental changes in real time.

AIThe vLLM Semantic Router team has released Decision 2.0, which answers multiple questions about one input in a single forward pass and outputs per-option probabilities. The post presents this as useful for routing and classification. A quoted post from Xunzhuo Liu says Decision 2.0 includes six open decision models ranging from 0.6B to 27B parameters, each topping same-size open models on the Jev Decision Index 0.3.

AIGoogle released EmbeddingGemma 2, a 740M-parameter multimodal embedding model under Apache 2.0 for private, on-device search and retrieval. It maps text, code, images, video, and audio into one shared space and reports a 9.92-point gain over EmbeddingGemma 1 on MTEB Code. The post lists about 191MB active RAM for quantized text-only weights and about 567MB for the full multimodal model on a Pixel 11 Pro.

AINACE introduces NDI 1.0, a document processing model for coding agents that it says is 90% cheaper and ranks first on the Parse Index. The company says it offers native MCP, SDK, and CLI integration for Claude Code, Codex, Hermes, OpenClaw, and PI, and supports 50 languages. NACE also states the model was trained on over 15M financial files and offers $25 in free API credits to developers.

AIArena (@arena) shared a link to its full Agent Arena overall leaderboard at arena.ai. The post provides no scores, rankings, or other details beyond directing readers to the leaderboard page.

AIShanghai AI Lab has open-sourced Intern-Decision, a family of 0.8B, 2B and 4B parameter models that return structured decisions with probabilities instead of free text. The developers self-report that the 4B model averages 90.02% accuracy across seven test suites, ahead of a commercial reference model at 88.74%, with about 44 milliseconds of local latency on a single RTX 4090. Weights are on Hugging Face, and MetaX says the models run on its hardware from launch.

AIOpenAI announced that GPT-6 is being rolled out to all ChatGPT users, alongside a new Intelligent UI. The excerpt provides no further details on capabilities, availability timing, or how the interface works.

AIAnthropic has released Claude Haiku 5.5, priced at one tenth of the previous Haiku 4.5 and one twentieth of Sonnet 5.5. The source excerpt is cut off before listing the full comparison, so the comparison with OpenAI's pricing is not confirmed here.

AIAnthropic released Claude Haiku 5.5, which the article says outperforms DeepSeek V4.1 Flash and GLM-5.3-Flash on official benchmarks and matches GPT-6 Luna on price. On OSWorld 2.1, its Low effort tier scores 42.0% at $0.07 per task, versus 15.7% at $1.45 for Haiku 4.5 at Max. Migrating from Haiku 4.5 requires changes to thinking configuration, sampling parameters, assistant prefill, and the computer-use tool version.

AIAnthropic released Claude Haiku 5.5, a small model priced at about a quarter of Haiku 4.5's running cost. Anthropic also halved Sonnet 5.5 cache read prices from 20 cents to 10 cents per million tokens.

AIXiaomi released the MiMo-V2.5-TTS Series, three speech synthesis models for stock voices, voice design, and voice cloning. The models accept natural-language style instructions and inline audio tags, and the source says the three models are free of charge for a limited time on the Xiaomi MiMo API platform. Xiaomi also open-sourced integration Skills for agent applications on GitHub.

Why it matters: The release shows how a TTS family adds style instructions, inline audio tags, and voice design or cloning to speech synthesis, which matters for agent and creative workflows.

AIXiaomi MiMo has released MiMo-V2.5-ASR, an open-source speech recognition model that the company says achieves state-of-the-art results across multiple benchmarks. The model supports bilingual Chinese–English recognition, Chinese dialects such as Wu, Cantonese, Hokkien, and Sichuanese, code-switching, and lyrics transcription. It is also designed to handle noisy environments and multi-speaker conversations.

AIXiaomi introduces the MiMo-V2.6 series of models, as stated on its official MiMo page. The source excerpt gives only the tagline, describing frontier intelligence across all modalities, so specific capabilities, benchmarks, and availability cannot be confirmed from it.

AIMiniMax says its open-source H3 model is almost on par with closed-source state-of-the-art video world models on physics. The claim is supported by a quoted benchmark, World Models' Last Exam in Physics, where eight leading models scored at most 57.76/100 across 40 physics tasks, and free-fall videos averaged only 26.61/100 on composite scores.

AIArtificial Analysis shares the second part of an AA-Video-T2V v2.0 prompt describing a four-shot documentary-style handheld video of a glass cabin in falling snow. The shots follow a caretaker sweeping snow from the deck, empty snow-covered windows, an empty interior, and the same caretaker stamping snow off his boots at the door, with hard cuts between shots.

AIArtificial Analysis reports that Grok Imagine Video 1.5 Lite comes closest to the frontier in Architecture & Real Estate, Consumer, and Productivity & Knowledge Work use cases. It sits furthest from the frontier in Live-Action Film and Frontier use cases. Against Grok Imagine Video 1.5, Lite matches it in Social Media & Creator Content and trails it on the other nine use cases.

AIArtificial Analysis shares an AA-Video-T2V v2.0 text-to-video prompt, a 10-second product-page scene of a presenter revealing a volcano-shaped mist diffuser. The prompt specifies a shift from warm daylight to dim evening lamplight, fixed close-up camera angles, and consistent presenter and product appearance across shots.

AIAmong 12 models on AA-Video-T2V-Silent v2.0, Grok Imagine Video 1.5 Lite is the only one that is both fastest and highest quality, with no model beating it on both measures. It generates a 10-second 1080p clip in a median of 60.5 seconds. Kling 3.0 1080p (Pro) scores slightly higher but takes 94 seconds for a 5-second clip, while Vidu Q3 Turbo is 9 seconds faster on a 5-second 720p clip yet scores well below it.

AIGrok Imagine Video 1.5 Lite ranks #17 on AA-Video-T2V v2.0, two places above Google's Veo 3.1. At 1080p with audio, it costs $0.14 per second versus $0.40 per second for Veo 3.1. Compared with Grok Imagine Video 1.5, Lite is 44% cheaper at 1080p but ranks six places lower.

AISpaceXAI's Grok Imagine Video 1.5 Lite ranks #17 on both AA-Video-T2V v2.0 leaderboards, ahead of Google's Veo 3.1 at about a third of its price. It is the fastest model at its quality level in Artificial Analysis benchmarks, with a median of 60.5 seconds for a 10-second 1080p clip, and it costs $0.14 per second at 1080p, 56% of Grok Imagine Video 1.5's $0.25 per second.

AIEpoch AI reports that OpenAI halved GPT-6.1 Sol's cached-input price compared with GPT-6 Sol. Its measurements also show the model handles long prompts faster. The post raises, but does not confirm, a new architecture for GPT-6.1 Sol.

AIAnthropic has cut Sonnet 5.5 cache read pricing in half to $0.10 per million tokens, with input at $2 and output at $10 per million tokens. The company says this makes Sonnet 5.5 roughly 20% cheaper on most agentic work. The change applies to API usage only and does not alter Claude Code usage limits.

AIAnthropic's Claude Haiku 5.5 reached 1587 points on the Arena WebDev leaderboard, a 257-point gain over Claude Haiku 4.5's 1330. The post says it also improved significantly across all categories.

AIGPT-6 Sol (Daybreak Blue, max) has been added to the Artificial Analysis Cyber Index as a trusted-access model and ranks #1 on the Index. Compared with the publicly available GPT-6 Sol, it shows its largest gains on CyberGym-E2E, the benchmark where the most safety refusals are observed.

AIArtificial Analysis added trusted-access models to its Cyber Index, and GPT-6 Sol (Daybreak Blue, max) now ranks first. The model is available only through OpenAI's Daybreak program and records no safety blocks across the Index. Its overall score is 32 points higher than the publicly available GPT-6 Sol (max), at a cost of $1.77 per task versus $11.67 for Grok 4.7 (xhigh).

AIThe Arena Alignment Index, built from over 90K real-world agent sessions across 27 models, ranks OpenAI's GPT-6.1-Sol first with a score of 87.9, ahead of Claude-Opus-5.5 at 83.2 and Grok-4.7 at 82.7. GPT-6.1-Sol also posted the lowest observed rates across the index's three signals: 0.89% Unauthorized Action, 1.98% False Attribution, and 2.34% Deceptive Completion. The index's authors report that newer models consistently outperform their predecessors across all four labs, suggesting broad progress in agent safety.

AIOpenAI's API pricing for GPT-6.1 Sol in Ultrafast mode is $12 per million input tokens and $60 per million output tokens. The mode is built for time-sensitive work such as debugging outages, agents navigating apps, and live experiences where speed matters.

AILightOn has released LightOnOCR-3, a family of OCR models in 0.8B and 4B versions that it says lead benchmarks including OlmOCR-Bench and ParseBench, with the 0.8B model positioned as the sub-1B option. The models recognize text, handwriting, images, charts and document structure in one pass, process documents up to twice as fast as LightOnOCR-2, and are released under the Apache 2.0 license.

AIOdyssey has launched Odyssey-3, its most powerful foundation world model, with a public research preview. Odyssey-3 Pro scored 66.1 on Physics-IQ Verified video-to-video with best-of-8 sampling, the highest reported result. The model generates environments from prompts and predicts changes in real time as users move through scenes.

AIOdyssey released Odyssey-3 Pro, which sets a new top score on Physics-IQ Verified, a benchmark where models continue videos of real physics experiments. In robotics, a robot arm with tens of hours of demonstrations recovered from a missed grasp, a behavior absent from those demos.

AIGeneralist has announced GEN-1.5, its latest foundation model for the physical world. The post provides only a link to the company's blog for further details, so no specifications, benchmarks, or availability information can be confirmed from this source.