Tulu Models by Allen Institute for Artificial Intelligence (AI2)
Last refreshed 2026-10-12. Next refresh: weekly.
Details
Links
WebsiteAbout
Tülu is Ai2 (Allen Institute for AI)’s open instruction-following model family built by post-training pretrained bases (notably Llama 3) with fully published data, code, and recipes. Tülu 3 uses a four-stage recipe—prompt curation, supervised fine-tuning, preference tuning, and Reinforcement Learning with Verifiable Rewards (RLVR)—aimed at knowledge, reasoning, math, coding, instruction following, chat, and safety. The family includes open instruct models across sizes, with a later Tülu 3 405B release applying the same recipe at 405B scale.
Decision facts
- Best fit
- codingmath-heavy prompts
- Capability starting point
- Tulu Llama 2 13B with 4K context
- Lowest tracked input
- Not tracked
- Closest related family
- OLMo
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
Use when the workload needs 4K context and 13B parameters.
Use when the workload needs 4K context and 7B parameters.
Use when the workload needs 4K context and 70B parameters.
Use when the workload needs 4K context and 13B parameters.
Use when the workload needs 4K context and 30B parameters.
Use when the workload needs 4K context and 65B parameters.
Use when the workload needs 4K context and 7B parameters.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Tulu Llama 2 13B | Use when the workload needs 4K context and 13B parameters. | 2023-11 | 4K context13B parameters | Current |
| Tulu Llama 2 7B | Use when the workload needs 4K context and 7B parameters. | 2023-11 | 4K context7B parameters | Current |
| Tulu Llama 2 70B | Use when the workload needs 4K context and 70B parameters. | 2023-11 | 4K context70B parameters | Current |
| Tulu 13B | Use when the workload needs 4K context and 13B parameters. | 2023-06 | 4K context13B parameters | Current |
| Tulu 30B | Use when the workload needs 4K context and 30B parameters. | 2023-06 | 4K context30B parameters | Current |
| Tulu 65B | Use when the workload needs 4K context and 65B parameters. | 2023-06 | 4K context65B parameters | Current |
| Tulu 7B | Use when the workload needs 4K context and 7B parameters. | 2023-06 | 4K context7B parameters | Current |
Release Timeline
2 release groupsSpecifications(7 models)
| Model | Released | Context | Parameters |
|---|---|---|---|
| Tulu Llama 2 13B | 2023-11 | 4K | 13B |
| Tulu Llama 2 7B | 2023-11 | 4K | 7B |
| Tulu Llama 2 70B | 2023-11 | 4K | 70B |
| Tulu 13B | 2023-06 | 4K | 13B |
| Tulu 30B | 2023-06 | 4K | 30B |
| Tulu 65B | 2023-06 | 4K | 65B |
| Tulu 7B | 2023-06 | 4K | 7B |
