Today, we’re introducing Ideogram's first vector model. Generate clean, editable SVGs with smooth gradients and sharp typography from a text prompt, or turn any image into a detailed, scalable vector. Live in Ideogram and the API: https://t.co/Cou2mo5p37 https://t.co/TWBuLiJ31a
#Model releases#Image generation#Multimodal
Ai2 has released the OlmoEarth v1.3 satellite embedding model, which generates 128-dimensional vectors for every 10-meter pixel worldwide, with the first 64 dimensions usable on their own, and has made public the model weights, pretraining data, and code. In 27 evaluations covering six continents, the 128-dimensional version scored highest on 20, and the 64-dimensional version led on 18, compared against AlphaEarth Foundations and Tessera v2. At the same time, the login-free OlmoEarth Embeddings Explorer has launched, allowing users to browse global embeddings by year, find similar locations, and train small classifiers on the spot with a few labeled points; data for 2025, 2024, and 2023 has global coverage, while 2017 to 2022 covers only some countries.
#Model releases#Open-source ecosystem#English text is being prepared.
#Model releases#Open-source ecosystem#Reasoning
The Allen Institute for AI released OlmoEarth v1.3 on October 22, 2026, including the OlmoEarth v1.3 Base model, 128-dimensional and 64-dimensional global precomputed annual embeddings, the login-free browser-based OlmoEarth Embeddings Explorer, and a technical report, all under the OlmoEarth Artifact License. The model is a 128-million-parameter ViT-Base foundation model that processes Sentinel-1, Sentinel-2, and Landsat image time series, outputting a 128-dimensional vector for every 10-meter pixel worldwide. Ai2 states that across 27 evaluations covering six continents, the 128-dimensional embeddings scored highest on 20 of them, while the 64-dimensional version led on 18; it also introduced PLANTEUR, a benchmark for crop types in tropical farmland, where the 128-dimensional embeddings achieved 24.9 mIoU in the benchmark's leave-one-island-out test, higher than Tessera v2's 19.1 and AlphaEarth Foundations' 15.8.
#Model releases#Open-source ecosystem#English text is being prepared.
Ai2 has released the satellite map embedding model OlmoEarth v1.3 and precomputed annual embedding data for the globe at multiple resolutions, while also launching the browser-based mapping tool OlmoEarth Embeddings Explorer, which can be used without logging in. The model compresses a year of satellite observations into a numerical fingerprint for each 10×10-meter patch of ground, and at the finest resolution it also offers two additional tiers of 64 square meters and 128 meters, which can be used to distinguish land cover such as farmland, forest, and mining areas. Ai2 says v1.3's training compute time is only 53% of v1.2's and 18% of v1's, with carbon emissions at 50% and 22% of the two, respectively.
#Model releases#Open-source ecosystem#English text is being prepared.
AI startup Recursive has released its first language model, Flint, with its Eureka Machine system involved in almost every step of the R&D process, as the system itself proposed ideas, implemented them, tested them, and planned subsequent experiments. Flint is a MoE model for agentic coding on local hardware, with 40B total parameters and 2.5B activated per token. It was completed in four months, and its weights are released on Hugging Face under the Apache 2.0 license, supporting llama.cpp, Ollama, LM Studio, vLLM, and SGLang. Recursive claims that it leads MoE models with fewer than 50B parameters on SWE-Bench Pro v2, Terminal-Bench 2.1, and SWE-Bench Verified, but the results have not yet been independently verified.
#Model releases#Open-source ecosystem#Agents
At the 2026 Global 1024 Developer Festival, iFlytek released the device-cloud integrated Spark X2.6 series models and open-sourced the Spark X2.6-4B on-device large model, raising device-cloud collaboration efficiency to 96.1%. Spark X2.6 is now available for trial on the iFlytek Spark official website, with input priced at a limited-time 0.9 yuan per million tokens and output at a limited-time 3.6 yuan per million tokens. The same event also saw the release of the SparkAudio 1.0 speech foundation large model, the Spark Smart Cockpit SparkAuto Space 3.0, and the Xingfan WinWork digital employee platform.
#Model releases#Agents#English text is being prepared.
#Model releases#Open-source ecosystem#English text is being prepared.
Ant Group Digital Technologies has released the decision model LingDT-Decision-1.0, offering two versions, 4B and 27B. It is now live on the Ant Group Digital Technologies DTMaaS platform with a limited-time free trial. According to the company, in scenarios such as customer service, risk control, and finance, it can reduce Token consumption by over 90% and improve response time by 10x compared with large language models. The model supports native multiple-choice and sequential decision-making (DAG), and its overall score on 23 open-source evaluation benchmarks surpasses decision models such as Microsoft-Decision-1 and Jev 1.13.0; the 27B version has the highest accuracy among models of the same parameter size.
#Model releases#Reasoning#English text is being prepared.
iFlytek releases the iFlytek Spark X2.6 cloud-device integrated large model, completing a capability upgrade based on fully domestic computing power. It includes the cloud-side 293-A30B model and the on-device X2.6-4B model, the latter of which has been open-sourced. The cloud model supports a one-million-token context, with an input price of 0.9 yuan per million tokens and an output price of 3.6 yuan per million tokens. iFlytek says its cloud-device routing accuracy has risen from 81.2% to 96.1%, and the inference efficiency of the domestic Feixing-2 chip has reached 4.5 times that before optimization.
#Model releases#English text is being prepared.#Speech
Google's next-generation model Gemini 4 Carbon has begun blind testing on LMArena under the name gemini-4-argon, and partners have also received test versions. Testers say its 3D scene generation quality is noticeably higher than Argon's, but the same task takes 2 to 4 times longer, while SVG and front-end design performance remains mediocre. Business Insider previously reported that Google employees who tried it internally felt that its coding felt very much like Opus 5.5.
#Model releases#Google#Evaluations / benchmarks
Recursive, founded by Tian Yuandong, has open-sourced its first model Flint, focused on local Agent coding, with 40 billion total parameters, 2.5 billion activated per token, and support for a 512k token context window. The company says that paired with its self-developed inference engine rec, Flint can reach at least 125 tokens/s on a MacBook Pro and over 150 tokens/s on RTX Spark, roughly 1.4–2.6 times higher throughput than llama.cpp. Flint scores 74.2% on SWE-bench Verified, 89.7% on SWE-bench Pro v2, and 54.9% on Terminal-Bench 2.1; rec currently supports M4 Pro, M5 Pro, and M5 Max, requiring at least 48GB of unified memory and about 30GB of free disk space.
#Model releases#Open-source ecosystem#Agents
At the main forum of the 2026 Global 1024 Developer Festival, iFlytek released Spark X2.6. Built on a fully domestic, self-reliant technology stack, it uses cloud-device collaboration to let on-device devices handle some tasks while complex problems are handed to the cloud, reducing Token costs and protecting sensitive data. Spark X2.6 is the first to achieve breakthroughs in coding and Agent capabilities based on fully domestic computing power, with the on-device Spark X2.6-4B reaching 41.9% coverage of Agent tasks. iFlytek says dynamic parallel training for long-horizon tasks improves ultra-long-context training efficiency by about 4x and inference efficiency on domestic clusters by 4.5x, and it has already been deployed in education, healthcare, legal and other fields.
#Model releases#English text is being prepared.#Agents
On October 22, iFlytek released the cloud-device integrated iFlytek Spark X2.6 at the 1024 Developer Festival, and simultaneously launched the on-device model Spark X2.6-4B; the system routes tasks to pure on-device, pure cloud, or cloud-device collaborative execution. Spark X2.6 uses a MoE architecture with 293B total parameters and 30B activated parameters, with the context window expanded to 1 million Tokens. The official website shows X2.6 at a limited-time 40% discount, with input and output prices of 0.9 yuan and 3.6 yuan per million Tokens, respectively. Liu Qingfeng expects that in the future about 60% of scenarios can be completed on-device, 30% will use cloud-device collaboration, and 10% will require cloud large models. The company says the share of tasks that can be completed independently on-device has risen from 12.8% to more than 40%.
#Model releases#Agents#English text is being prepared.
Hugging Face on Thursday released the open-source DNA understanding model Carbon-X, which can run locally on a laptop, letting users ask in natural language how DNA changes affect gene activity and health. The model was developed in collaboration with researchers at Zhongguancun Academy and the University of Naples Federico II; it extends the vocabulary of an existing language model to read DNA in six-letter units and was further trained on 300 billion tokens of DNA sequences, English text, and samples pairing DNA with biological descriptions. The model has 14B parameters and can process sequences of about 50,000 base pairs at a time; it cannot upload a complete human genome. The developers also note that the prediction accuracy has not yet been validated through wet-lab experiments.
#Model releases#Open-source ecosystem#Hugging Face
Hugging Face has released the open-source model Carbon-X, combining DNA analysis and generation with natural language so that researchers can ask questions directly and receive predictions and explanations. The model is based on Qwen3-14B and was trained on genomic sequences and natural language from humans, other mammals, plants, and fungi, with viruses explicitly excluded from the training data. Researchers can run it locally, avoiding the sharing of sensitive patient data with external AI companies, and can also generate new DNA sequences to assist gene therapy research.
#Model releases#Open-source ecosystem#English text is being prepared.
Ant Digital Technologies has released its decision model LingDT-Decision-1.0, purpose-built for rapid industrial decision-making, offering 4B and 27B versions. It is now live on the Ant Digital Technologies DTMaaS platform with a limited-time free trial. The model natively supports multiple-choice and sequential decision-making (DAG), and outperforms decision models such as Microsoft-Decision-1 and Jev 1.13 across 23 open-source evaluation benchmarks, with the 27B version achieving the highest accuracy among models of the same parameter size. Tests by Ant Digital Technologies in customer service, risk control, finance and other scenarios show that, when handling the same tasks, it can reduce Token consumption by over 90% compared with large language models and improve response time by 10x.
#Model releases#Reasoning#English text is being prepared.
English text is being prepared.
#Model releases#Open-source ecosystem#Agents
The Xingchen General Artificial Intelligence Laboratory of China Telecom Artificial Intelligence Technology Co., Ltd. has released and open-sourced the dedicated intent understanding model Xing4.0-29B-Intent, simultaneously opening 2.753 million training corpus entries, model weights, and inference images. The model is fine-tuned based on Xing4.0-29B-A4B and adopts a unified schema-driven semantic parsing framework, supporting both intent recognition and parameter extraction. It outputs field-constrained JSON, with an average output of a few dozen tokens, and the full result response latency can reach the hundreds-of-milliseconds level. The model has been applied in conversational intent recognition and intelligent ticket routing for Telecom's ten-thousand-line customer service, as well as spoken command parsing for smart home speakers. The code is open under the Apache 2.0 license, and the dataset and weights will be released simultaneously on Hugging Face.
#Model releases#Open-source ecosystem#Reasoning
IBM has released the open-source speech language model Granite Speech 5.0 2B under the Apache 2.0 license. It supports speech-to-text in 6 languages and also offers speech translation capabilities. On the Open ASR Leaderboard, the model ranks second among open-weight models with an average English WER of 5.68%, behind only Qwen3-ASR-1.7B at 5.40%. Compared with 4.1 2B, it adds language ID prompting, keyword biasing of up to 2000 words, and on-demand punctuation and capitalization, and can run on Transformers, vLLM, llama.cpp, and mlx-audio.
#Model releases#Speech#Open-source ecosystem