USASI
Model family

Tülu

Family overview — summarizes releases; licenses and availability belong to each release.

Maintained by Ai2 (Allen Institute for AI)15

Tülu is Ai2's line of instruction-following models together with the open post-training recipe used to produce them. Tülu 3 (November 2024) applied supervised fine-tuning, Direct Preference Optimization, and reinforcement learning with verifiable rewards (RLVR) to Meta's Llama 3.1 base models at 8B and 70B, followed by a 405B model in January 2025 and by Tülu 3.1 8B, which changed only the final RL stage (from PPO to GRPO).12345

Last reviewedEntry updated Documented release Nov 21, 2024

Releases assessed

Each release is assessed on its own. This family has no family-wide openness label.
Releases in the Tülu family
ReleaseWeightsTier (USASI rubric v0.1)LicenseReleased
Llama 3.1 Tülu 3.1 8BWeights: PublicModel-disclosure tier (USASI rubric v0.1): Open-stackLlama 3.1 Community License Agreement, Apache-2.0Unknown

What it is useful for

Ai2 presents Tülu 3 as a documented, reproducible post-training package, releasing the models together with training data, code, and evaluation tools. The model card notes it is one step in the process of training fully open models such as OLMo.125

Organization context

Provenance and derivatives

The Tülu 3 models are post-trained from Meta's Llama 3.1 base models (8B, 70B, and 405B); the Tülu cards state that the size and composition of the Llama 3.1 pretraining corpus is unknown. Ai2 supplies the post-training data (the Tülu 3 SFT mixture, preference mixtures, and RLVR prompts), code, and recipe.265

U.S. eligibility

Project eligibility rests on documented governing or maintaining entities, not on contributors.

Eligible · basis: U.S. nonprofit or lab

The Tülu models and recipe are developed and released by Ai2 (Allen Institute for AI), which describes itself as a Seattle-based non-profit AI research institute. The Tülu 3 models are fine-tunes of Meta's Llama 3.1 base models; the base models are recorded as provenance. Eligibility of each release is assessed on the release record.157

Assessed Sep 29, 2026

Sources

  1. 1.
    Tulu | Ai2 (external site: allenai.org)

    Ai2 · Official page · accessed Sep 29, 2026

  2. 2.
  3. 3.
    Tülu 3: The next era in open post-training (external site: allenai.org)

    Ai2 · Announcement · published Nov 21, 2024 · accessed Sep 29, 2026

  4. 4.
  5. 5.
  6. 6.
  7. 7.
    About us | Ai2 (external site: allenai.org)

    Ai2 · Official page · accessed Sep 29, 2026

This listing is not an endorsement, a safety assessment, or a federal approval.

Support Us

Help keep USASI useful.

Optional. No USASI account required. Payment takes place on the linked provider’s website (Buy Me a Coffee).

About supporting this project