Pins matching company:IBM

Ted Codd's paper in Communications of the ACM proposes storing data as relations queried by predicate logic, independent of how it is laid out on disk. It is the founding document of the relational database and of SQL.

At its TechXchange event IBM releases Granite 3.0, its third generation of open Apache 2.0 models: 8B and 2B dense language models, Granite Guardian safety models and small mixture-of-experts models.

IBM releases Granite 3.1: every Granite 3 model gains a 128K-token context window, with new Granite Embedding models and function-calling hallucination detection in Granite Guardian 3.1, all under Apache 2.0.

IBM releases Granite 3.2 with switchable chain-of-thought reasoning in the 8B and 2B Instruct models and its first official vision-language model, Granite Vision 3.2 2B, all under Apache 2.0.

IBM releases Granite 3.3 8B and 2B Instruct with fill-in-the-middle support and refined reasoning, plus Granite Speech 3.3 8B, its first official speech-to-text model, under Apache 2.0.

IBM releases Granite 4.0 Tiny Preview, an early look at a 7B-parameter hybrid Mamba-2/transformer mixture-of-experts model with 1B active parameters, trained on only 2.5 trillion tokens so far.

IBM releases Granite 4.0, open Apache 2.0 language models built on a hybrid Mamba-2/transformer architecture that IBM says cuts memory needs by over 70%, the first open models with ISO 42001 certification.

IBM releases Granite 4.0 Nano, its smallest models yet: hybrid and plain-transformer versions of roughly 1B and 350M parameters for edge and on-device use, under Apache 2.0.

IBM releases Granite 4.1, dense 3B, 8B and 30B language models under Apache 2.0 whose 8B instruct matches or beats the 32B-parameter Granite 4.0 mixture-of-experts model, alongside new vision, speech, guardian and embedding models.

IBM releases Granite 4.2, three open Apache 2.0 reasoning models (3B, 8B and 30B) with a switchable thinking mode; the 8B scores 47.67 on SWE-bench Verified and 86.67 on AIME25 in IBM's own runs.