Ai2 logo

Ai2 models

The Allen Institute for AI is a nonprofit research institute that releases fully open models such as OLMo, OLMoE, Tülu and Molmo together with their training data, code and checkpoints.

Seattle, United StatesFounded 2014allenai.org

Releases
23
Open weights
23
Latest
Oct 02, 2026

October 2026

1 release
  1. LLMOpen weights

    AstaBrief 8B

    Open-weights model that turns a research question and retrieved literature excerpts into a cited scientific report, powering Fast mode in Asta's report generation.

    Modalities
    text → text
    Parameters
    8B
    License
    Apache-2.0
    At launch
    Open weights
    • Post-trained from Qwen3-8B
    • Training data open-sourced with the model
    • Released under Apache 2.0
    Source: Open-sourcing AstaBrief, the fast report-generation model in Asta | Ai2

May 2026

2 releases
  1. LLMOpen weights

    EMO

    Research mixture-of-experts model pretrained so that modular structure emerges from data, letting a task use about 12.5% of its experts with near full performance.

    Modalities
    text → text
    Parameters
    MoE: 14B total, 1B active
    License
    Apache-2.0
    At launch
    Open weights
    • 128 experts in total with 8 active
    • Trained on 1 trillion tokens
    • Released under Apache 2.0
    Source: EMO: Pretraining mixture of experts for emergent modularity | Ai2
  2. MultimodalOpen weights

    MolmoAct 2

    Upgrade to Ai2's Action Reasoning Model that pairs the Molmo 2-ER embodied reasoning model with a flow-matching action expert for real-world robots.

    Modalities
    text, image → action
    At launch
    Open weights
    • Molmo 2-ER released as the reasoning backbone
    • Includes an open action tokenizer
    • Full code and training data released later
    Source: MolmoAct 2: An open foundation for robots that work in the real world | Ai2

March 2026

2 releases
  1. MultimodalOpen weights

    MolmoWeb

    Open visual web agent built on Molmo 2 that navigates websites, fills out forms and retrieves information in a browser, designed for self-hosting.

    Modalities
    text, image → text
    Parameters
    4B and 8B
    License
    Apache-2.0
    At launch
    Open weights
    • Released with weights, training data and evaluation tools
    • Training data includes 36K human task trajectories
    • Released under Apache 2.0
    Source: MolmoWeb: An open agent for automating web tasks | Ai2
  2. LLMOpen weights

    Olmo Hybrid

    Fully open 7B model family that mixes transformer attention with linear recurrent layers, released with a controlled comparison against Olmo 3 7B.

    Modalities
    text → text
    Parameters
    7B
    License
    Apache-2.0
    At launch
    Open weights
    • Combines transformers and linear RNNs
    • Developed through 1B and 7B scale experiments
    • Released under Apache 2.0
    Source: Introducing Olmo Hybrid: Combining transformers and linear RNNs for superior scaling | Ai2

January 2026

1 release
  1. CodingOpen weights

    SERA

    First release in Ai2's Open Coding Agents family, coding models built on Qwen 3 with a training method for adapting agents to any repository.

    Modalities
    text → text
    Parameters
    8B to 32B
    License
    Apache-2.0
    At launch
    Open weights
    • Trained up to 32K context length
    • SERA-14B added on February 3, 2026
    • Released under Apache 2.0
    Source: Open Coding Agents: Fast, accessible coding agents that adapt to any repo | Ai2

December 2025

3 releases
  1. LLMOpen weights

    Bolmo

    Family of byte-level language models created by byteifying Olmo 3 models, reusing their backbones while replacing subword tokenization with byte-level patching.

    Modalities
    text → text
    Parameters
    1B and 7B
    License
    Apache-2.0
    At launch
    Open weights
    • Uses an mLSTM local encoder and a boundary predictor
    • Built on the public Olmo 3 7B checkpoint
    • Released under Apache 2.0
    Source: Introducing Bolmo: Byteifying the next generation of language models | Ai2
  2. ReasoningOpen weights

    Olmo 3.1

    Update to Olmo 3 with two new 32B checkpoints, Olmo 3.1 Think 32B and Olmo 3.1 Instruct 32B, plus refreshed RL Zero 7B models.

    Modalities
    text → text
    Parameters
    32B
    License
    Apache-2.0
    At launch
    Open weights
    • Think 32B came from 21 more days of RL training on 224 GPUs
    • Instruct 32B applies the Olmo 3 Instruct 7B recipe at larger scale
    • Released under Apache 2.0
    Source: Olmo 3: Charting a path through the model flow to lead open-source AI | Ai2
  3. MultimodalOpen weights

    Molmo 2

    Successor to Ai2's Molmo that extends its grounded vision capabilities to video and multi-image understanding, adding video pointing and tracking, released with open weights.

    Modalities
    text, image, video → text
    Parameters
    4B, 7B and 8B
    License
    Apache-2.0
    At launch
    Open weights
    • Molmo 2 8B and 4B are built on Qwen 3
    • Molmo 2-O 7B is built on Olmo for a fully open stack
    • Full training codebase released later
    Source: Molmo 2: State-of-the-art video understanding, pointing, and tracking | Ai2

November 2025

1 release
  1. ReasoningOpen weights

    Olmo 3

    Third generation of fully open Olmo models, released as Base, Think, Instruct and RL Zero variants with checkpoints from every major stage of training.

    Modalities
    text → text
    Parameters
    7B and 32B
    License
    Apache-2.0
    At launch
    Open weights
    • Pretrained on the new Dolma 3 data
    • Releases base, mid-trained, long-context and post-training checkpoints
    • Released under Apache 2.0
    Source: Olmo 3: Charting a path through the model flow to lead open-source AI | Ai2

October 2025

1 release
  1. MultimodalOpen weights

    olmOCR 2

    Updated OCR model for English digitized print documents, built on Qwen2.5-VL-7B and trained with unit test rewards for document correctness.

    Modalities
    text, image → text
    Parameters
    7B
    License
    Apache-2.0
    At launch
    Open weights
    • Released as olmOCR-2-7B-1025
    • Fine-tuned on olmOCR-mix-1025 with 270,000 PDF pages
    • Released under Apache 2.0
    Source: olmOCR 2: Unit test rewards for document OCR | Ai2

August 2025

2 releases
  1. SpeechOpen weights

    OLMoASR

    Family of fully open English automatic speech recognition models from Ai2, trained from scratch on a curated, large-scale dataset and released in sizes from 244M to 1.5B.

    Modalities
    audio → text
    Parameters
    244M to 1.5B
    License
    Apache-2.0
    At launch
    Open weights
    • Sizes include small.en at 244M, medium.en at 769M and two 1.5B large.en models
    • The large.en-v2 model was trained on 680,000 hours of audio
    • Released under Apache 2.0
    Source: OLMoASR: A series of open speech recognition models | Ai2
  2. MultimodalOpen weights

    MolmoAct

    Ai2's first Action Reasoning Model, a robotics model that reasons about its environment in 3D before acting, released with an open post-training dataset.

    Modalities
    text, image → action
    Parameters
    7B
    License
    Apache-2.0
    At launch
    Open weights
    • Post-trained on about 10,000 distinct robot episodes
    • Pretrained on a curated subset of Open-X Embodiment data
    • Released under Apache 2.0
    Source: MolmoAct: An Action Reasoning Model that reasons in 3D space | Ai2

July 2025

1 release
  1. LLMOpen weights

    FlexOlmo

    Language model training paradigm in which each data owner branches from a shared public model and trains an expert locally on its own data, letting owners opt in or out later.

    Modalities
    text → text
    License
    Apache-2.0
    At launch
    Open weights
    • Enables collaboration without pooling sensitive data
    • Released as FlexOlmo-7x7B-1T on Hugging Face
    • Released under Apache 2.0
    Source: Introducing FlexOlmo: a new paradigm for language model training and data collaboration | Ai2

March 2025

1 release
  1. LLMOpen weights

    OLMo 2 32B

    Largest model in the OLMo 2 family and a fully open model trained on up to 6 trillion tokens and post-trained with Tülu 3.1.

    Modalities
    text → text
    Parameters
    32B
    License
    Apache-2.0
    At launch
    Open weights
    • Trained with the new OLMo-core codebase
    • Trained on Google Cloud's Augusta AI Hypercomputer
    • Released under Apache 2.0
    Source: OLMo 2 32B: First fully open model to outperform GPT 3.5 and GPT 4o mini | Ai2

February 2025

1 release
  1. MultimodalOpen weights

    olmOCR

    Open PDF text extraction toolkit built on a fine-tuned Qwen2-VL-7B-Instruct model, released with weights, fine-tuning data and an efficient inference pipeline.

    Modalities
    text, image → text
    Parameters
    7B
    License
    Apache-2.0
    At launch
    Open weights
    • Built on Qwen2-VL-7B-Instruct
    • Released with training and inference code
    • Followed by olmOCR 2 in October 2025
    Source: olmOCR: Efficient PDF text extraction with vision language models | Ai2

January 2025

1 release
  1. LLMOpen weights

    Tülu 3 405B

    Application of the fully open Tülu 3 post-training recipe to the Llama 3.1 405B base model, used to stress-test reinforcement learning with verifiable rewards at scale.

    Modalities
    text → text
    Parameters
    405B
    License
    Llama 3.1 Community License
    At launch
    Open weights
    • Trained with reinforcement learning from verifiable rewards
    • Served with vLLM using 16-way tensor parallelism during RL
    • Built on Llama 3.1 405B
    Source: Scaling the Tülu 3 post-training recipes to surpass the performance of DeepSeek V3 | Ai2

November 2024

2 releases
  1. LLMOpen weights

    OLMo 2

    Second generation of fully open OLMo models, trained on up to 5 trillion tokens and released with weights, data, code, recipes, checkpoints and instruct versions.

    Modalities
    text → text
    Parameters
    7B and 13B
    License
    Apache-2.0
    At launch
    Open weights
    • Released in 7B and 13B sizes
    • OLMo 2-Instruct models apply the Tülu 3 post-training recipe
    • Released under Apache 2.0
    Source: OLMo 2: The best fully open language model to date | Ai2
  2. LLMOpen weights

    Tülu 3

    Family of post-trained Llama 3.1 models released with all of their data, data mixes, recipes, code, infrastructure and evaluation framework.

    Modalities
    text → text
    Parameters
    8B and 70B
    License
    Llama 3.1 Community License
    At launch
    Open weights
    • Opens up the full post-training recipe
    • Built on Llama 3.1 base models
    • Later scaled to 405B parameters
    Source: Tülu 3 opens language model post-training up to more tasks and more people | Ai2

September 2024

2 releases
  1. MultimodalOpen weights

    Molmo

    Family of open vision-language models trained without distillation from proprietary VLMs, notable for learning to point at what they perceive in images.

    Modalities
    text, image → text
    Parameters
    1B (MolmoE), 7B and 72B
    License
    Apache-2.0
    At launch
    Open weights
    • Weights, captioning data and fine-tuning data released
    • Released as MolmoE-1B, Molmo-7B-O, Molmo-7B-D and Molmo-72B
    • Built from pretrained vision encoders and language-only LLMs
    Source: Molmo | Ai2
  2. LLMOpen weights

    OLMoE

    First mixture-of-experts model in the OLMo family, developed with Contextual AI and released with open data, code, logs and intermediate checkpoints.

    Modalities
    text → text
    Parameters
    MoE: 7B total, 1B active
    License
    Apache-2.0
    At launch
    Open weights
    • Trained on 5 trillion tokens
    • 244 pretraining checkpoints released, one every 5,000 steps
    • Instruct version previewed the Tülu 3 post-training pipeline
    Source: OLMoE: An open, small, and state-of-the-art mixture-of-experts model | Ai2

April 2024

1 release
  1. LLMOpen weights

    OLMo 1.7 7B

    Updated 7B OLMo model with a longer context window, a new two-stage training procedure and improved data, released with the Dolma 1.7 dataset.

    Modalities
    text → text
    Context
    4K context
    Parameters
    7B
    License
    Apache-2.0
    At launch
    Open weights
    • Context length raised from 2,048 to 4,096 tokens
    • Trained on the new Dolma 1.7 data mix
    • Released under Apache 2.0
    Source: OLMo 1.7–7B: A 24 point improvement on MMLU | Ai2

February 2024

1 release
  1. LLMOpen weights

    OLMo

    Ai2's first truly open language model, released with its full pretraining data, training code, model weights and evaluation tools so researchers can study how it was built.

    Modalities
    text → text
    Context
    2K context
    Parameters
    1B and 7B
    License
    Apache-2.0
    At launch
    Open weights
    • Pretrained on the open Dolma corpus of about three trillion tokens
    • Built with partners including the Kempner Institute, AMD and the LUMI supercomputer
    • Available on Hugging Face and GitHub
    Source: Hello OLMo: A truly open LLM | Ai2

Cite and reuse this data

These release records are free to reuse under CC BY 4.0, including commercially. Credit The AI Engineering and link to this page's URL.

Dates are the labs' official announcement dates, and every record links to its source. List fields in the CSV are separated by " | ".

Citation

The AI Engineering Editors. "Ai2 models: release history and timeline." The AI Engineering, updated Oct 02, 2026. https://theaiengineering.com/releases/ai2

HTML link

<a href="https://theaiengineering.com/releases/ai2">Ai2 models: release history and timeline</a> (The AI Engineering)
Full dataset
1,167 releases from 55 labs
Coverage
to
Updated
License
CC BY 4.0
Publisher
The AI Engineering

Other labs