Atlas

Models

← All models

Olmo 3 7B Think

Allen Institute for AI (Ai2) · 2025-11-20 · 7.3B parameters

Olmo 3 7B Think is Ai2's fully open 7B reasoning checkpoint, built from Olmo 3 Base through thinking SFT, DPO, and reinforcement learning with verifiable rewards. It generates intermediate thinking traces for math, coding, and general problem solving, and Ai2 publishes the weights, data, code, and checkpoints across the model flow.

Benchmark scores

BenchmarkScore
AA-LCR0.0
AA-Omniscience Index-73.7
AIME 202570.7
Artificial Analysis Intelligence Index4.0
Artificial Analysis Omniscience Accuracy10.8
Artificial Analysis Omniscience Hallucination Rate94.7
Artificial Analysis Openness Index88.9
Capability129.4
CritPt0.0
GPQA Diamond51.6
Humanity's Last Exam (Text-Only)5.7
IFBench (Artificial Analysis)41.5
LiveCodeBench (Artificial Analysis)61.7
MMLU Pro65.5
SciCode (Artificial Analysis)21.2
SnakeBench Average Apples6.8
SnakeBench Best Apples21.0
SnakeBench Rating25.2
SnakeBench Total Apples81.0
SnakeBench Win Rate66.7
Terminal-Bench Hard0.8
τ²-bench (Telecom)0.0
Loading Atlas data…