IBM Granite
IBM has (rather quietly) released updates to its Granite series by continuing post-training on its 3B dense, 8B dense, and 30B dense models. In terms of raw performance, however, these models are behind the likes of Qwen and Gemma.
A time-series model for forecasting.
A small speech-to-text model supporting six languages. It also supports the generation of English audio for translation.
Of course, IBM can't stop training tiny models and to add them to their model families. The new Granite models feature hybrid attention and MoE architecture first time.
We've been covering IBM and their Granite LLM series for a while. With this series, IBM finally scaled up the model size as well, bringing a series of hybrid (attention + mamba) models, ranging from a 3B dense to a 32B-A9B MoE. We used the models and were impressed, although not surprised, by the quality, given the continued persistence of IBM's team to release better and better models. Granite, for at least the 3B variant, is roughly in the SmolLM3 quality range, being only surpassed by Qwen3 4B in terms of multilingual and instruction following capabilities. The tone of Granite 4.0 is refreshingly non-exciting compared to the sloptimized models recently (i.e. the trend across the industry for playful, emoji-filled, and often sycophantic models), making it feel like old Mistral models in a good way. Interestingly enough, they are also following Qwens lead and will release a separate reasoning model later in the year. We've heard many reports from people training models that hybrid reasoning - i.e. a toggle of thinking tokens on and off - adds a major complexity cost in training that lowers the peak performance of both modes. IBM debuted the hybrid thinking approach (togglable via prompts) very early on for open models, which was adopted by others later.
As the documents yearn to be reranked after the retrieval step, IBM also releases an updated version alongside the embedding models.
IBM continues to release models for the whole stack of a RAG system, this time updating their embedding models while switching to the ModernBERT architecture.
A classifier for jailbreak attempts, RAG outputs, etc.
A speech-to-text model from the IBM team.
Avid readers of the Artifacts posts are not surprised by the regular appearances of IBM. They are back with a preview of their next generation of models, featuring a fine-grained MoE architecture.
An update to the Granite series by IBM. Alongside a toggle-able reasoning mode, which was previewed as an experiment, it also supports fill-in-the-middle to be used as an in-line coding model.
A safety classifier by IBM for text classification.
IBM is constantly churning out new (mostly small-ish) models, but is still a relatively unknown player among the model makers. This model is their first shot at a reasoning model and builds upon Granite 3.1 8B, which we covered in the previous episode. While a lot of reasoning models released recently rely on distillation from R1, IBM uses "own reinforcement learning-based techniques for triggering chain-of-thought reasoning across any domain", which does not need a teacher model as outlined in their blog. Furthermore, the reasoning mode of the model can be toggled on or off by setting a specific system message. It is obvious that the distinction between "normal" LLMs and "reasoning" LLMs will become non-existent in the future as RL becomes an increasingly big part of the training. Future models will learn when and how much inference should be spent before the final answer.
Another IBM preview model, with this being their first shot at a VLM. It combines its Granite language model with SigLIP. The presented numbers look promising, and, similar to other IBM models, are released under Apache 2.0.
Largely flying under the radar, IBM steadily releases models in their Granite series to rival the smaller Llama models under an Apache 2.0 license.