AI Potluck
Model components / Fine-tuned / chat models

OLMo 3 Instruct

Ai2

Instruction-tuned chat SKUs of Ai2's OLMo 3 family (7B/32B-Instruct), post-trained on the fully-open OLMo 3 base. OLMo 3-Instruct 7B ties or surpasses Qwen 2.5, Gemma 3 and Llama 3.1 at its scale. Apache-2.0; open weights, data, recipe and checkpoints. Supersedes OLMo 2 Instruct.

Apache-2.0; fully-open instruct line post-trained with the open Tulu/OLMo recipe.

Openness

5 high confidence
5.0
weights
open(Apache-2.0)
data
open(Dolma 3 pretraining + open post-training mixtures)
code
open(OLMo + open-instruct)
checkpoints
open
license
Apache-2.0(OSI)

Fully-open instruct line: Apache-2.0 weights + open pretraining and post-training data + open training code + checkpoints.

Adoption

3 medium confidence
3.0

Solid research/practitioner reach for a fully-open instruct line; below the Qwen/Llama instruct workhorses. Downloads still ramping as it supersedes OLMo 2 Instruct.

Capability

4 high confidence
4.0

Strongest fully-open chat line; competitive with same-scale open-weight instruct models but below the 2026 frontier (=5).

Unchanged since 2026-07-29 (last edited, not re-checked)