Total 84 · Clear filters
On 22 October 2024 Anthropic announced an upgraded Claude 3.5 Sonnet, a new Claude 3.5 Haiku and computer use in public beta, which lets Claude operate a computer through screenshots, cursor movements and typing.
On 19 September 2024 Alibaba released Qwen2.5 with 13 open-weight models. The family was trained on up to 18 trillion tokens and supports 128K tokens of context in more than 29 languages.
On 5 September 2024 DeepSeek merged its chat and coder models into DeepSeek-V2.5, which kept both conversational ability and code processing and went live on its web chat and API.
On 5 September 2024 01.AI released the open Yi-Coder code models in 1.5B and 9B sizes. The 9B chat model scored 23.4% on LiveCodeBench, above every other model under 10 billion parameters, according to the company.
On August 13, 2024, OpenAI released SWE-bench Verified, a 500-sample human-validated subset of SWE-bench built with the benchmark's original authors from Princeton, after filtering ambiguous tasks and unreliable tests.
On 24 July 2024 Mistral AI released Mistral Large 2, a 123-billion-parameter dense language model with a 128K context window, whose weights were published on Hugging Face under a research licence and which launched across several cloud platforms.
On 16 July 2024 Mistral AI released two specialised 7B open models: Codestral Mamba, built on the Mamba architecture for code, and Mathstral, a math model made with Project Numina.
On 20 June 2024 Anthropic released Claude 3.5 Sonnet, the first model of the 3.5 family, and launched Artifacts on claude.ai. It cost the same as Claude 3 Sonnet and ran twice as fast as Claude 3 Opus.
On 17 June 2024 DeepSeek released DeepSeek-Coder-V2, an open-source mixture-of-experts code model trained on about 6 trillion additional tokens, with a 128K context window.
On 29 May 2024 Mistral AI released Codestral, a 22-billion-parameter open-weight code model trained on more than 80 programming languages, with a 32K context window and weights under a non-production licence.
On 28 February 2024 BigCode and NVIDIA released StarCoder2 in 3B, 7B and 15B sizes. The new training set, The Stack v2, is 67.5 TB and is built from the Software Heritage archive.
FunSearch, published in Nature in December 2023, used large language models for program search rather than direct solution search, and found results better than the previous best on the cap set problem and online bin packing.
On 2 November 2023 DeepSeek released its first model, DeepSeek Coder, a set of open-source code models from 1.3B to 33B parameters trained on 2 trillion tokens each.
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 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.
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.
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 21 June 2022 GitHub made Copilot generally available to individual developers at 10 US dollars a month, after a preview that drew more than 1.2 million developers. It marked the arrival of an AI coding assistant inside mainstream editors.
On 2 February 2022, DeepMind introduced AlphaCode, a system that writes computer programs at a competitive level and, on average, ranked within the top 54% of participants across 10 Codeforces contests.
On 10 August 2021 OpenAI released an improved Codex, a GPT model fine-tuned on public code, through its API in private beta. The model powered GitHub Copilot and came with the new HumanEval benchmark.
On 29 June 2021 GitHub announced GitHub Copilot as a technical preview, an AI pair programmer that suggests lines and whole functions as developers write code. It was powered by OpenAI Codex.
