Total 671 · Clear filters
In August 2023 Alibaba released the Qwen-VL vision-language models, built on the Qwen language model. The paper was posted to arXiv on 24 August 2023, and the code, demos and models were made public.
On 24 August 2023 Meta released Code Llama, a family of code models built by further training Llama 2 on code. It came in 7B, 13B and 34B sizes, under the same community licence as Llama 2, and a 70B version followed in January 2024.
On 18 August 2023, AgiBot released the Yuanzheng A1, its first humanoid robot, at its first public product launch. The machine was pitched as a large-model-driven, embodied-intelligence body aimed first at manufacturing.
On 15 August 2023, Unitree released the H1, its first general-purpose humanoid robot, a roughly 180 cm, 47 kg machine that the company described as China's first full-size general-purpose humanoid able to run.
On 15 August 2023 iFlytek released SparkDesk V2.0, upgrading its large language model's coding and multimodal abilities and launching the iFlyCode programming assistant. The launch was followed by an 8.63% fall in the company's shares.
Alibaba's first open-weight Qwen releases: Qwen-7B appeared on 3 August 2023, followed by 14B, 1.8B and 72B models, giving outside developers downloadable models from small to large, with published training volumes, context limits and benchmark scores.
In August 2023 a ByteDance subsidiary opened its Doubao chatbot for invitation testing, built on a model called Skylark that was among the first LLMs to pass China's algorithm registry.
On 28 July 2023, Google DeepMind introduced RT-2, a vision-language-action model that carries knowledge learned from web data over to robot control and roughly doubled generalisation to unseen scenarios compared with its predecessor.
On 26 July 2023 Stability AI announced SDXL 1.0, its new flagship open image model, after a limited research-only release of SDXL 0.9; a one-step SDXL Turbo followed in November.
On 18 July 2023 Meta released Llama 2, a family of 7B, 13B and 70B parameter models free for research and commercial use under a custom license. Microsoft Azure, AWS and Hugging Face were among the platforms offering it.
On 11 July 2023 Anthropic released Claude 2 with better performance and longer responses, and opened the claude.ai beta to the public in the US and UK. Claude 2 accepted up to 100K tokens per prompt.
In July 2023 Baichuan, a Chinese start-up founded by Sogou founder Wang Xiaochuan, released Baichuan-13B as open models. Commercial use was allowed after registration.
On 20 June 2023 Microsoft Research posted the paper 'Textbooks Are All You Need' to arXiv, introducing phi-1, a 1.3-billion-parameter code model trained on textbook-quality data.
In late June 2023, Baidu announced ERNIE 3.5, an upgrade to ERNIE Bot that it said beat ChatGPT in overall ability and GPT-4 in some Chinese-language tasks, with higher throughput and a plugin architecture.
Baichuan, a Chinese AI startup, released Baichuan-7B as an open-source, bilingual Chinese-English language model about two months after its founding.
At Google I/O on 10 May 2023, Google introduced PaLM 2, a large language model in four sizes — Gecko, Otter, Bison and Unicorn — said to improve multilingual, reasoning and coding abilities and to power more than 25 products, including Bard.
On 6 May 2023 iFlytek unveiled SparkDesk, its answer to ChatGPT, aimed at education and enterprise uses. Chairman Liu Qingfeng claimed it already beat ChatGPT in several areas and promised parity in English by October 2023.
On 4 May 2023 the BigCode project, led by Hugging Face and ServiceNow, released StarCoder and StarCoderBase, two openly licensed code models of about 15.5 billion parameters trained on permissively licensed GitHub data.
On 2 May 2023 Inflection AI launched Pi, a chatbot it called personal intelligence and presented as a kind, supportive companion.
On 25 April 2023 Hugging Face released HuggingChat, a free chatbot it described as an open-source alternative to ChatGPT, running a LLaMA-based model built by the volunteer Open Assistant project.
On 19 April 2023 Stability AI released Stable LM, open language models at 3B and 7B parameters, with base versions under a share-alike license and research-only instruction-tuned versions.
