# GSM8K — AI model rankings

Grade-school maths word problems that take a few steps of arithmetic to work through. It separated the models of 2022 and 2023 sharply, then saturated. One caveat on the historical numbers: OpenAI included part of the GSM8K training set in GPT-4's pre-training mix, so GPT-4's score is not a clean few-shot result. Higher is better.

5 tracked models have a published GSM8K score. Higher is better. Scores come from published lab reports and benchmark sources; recorded sources appear beside the scores.

## Ranking

| Rank | Model | Developer | Score | Source | Released |
| --- | --- | --- | --- | --- | --- |
| 1 | Gemma 3 | Google | 95.9% | Lab | Mar 12 2025 |
| 2 | GPT-4 | OpenAI | 92% | Lab | Mar 14 2023 |
| 3 | DeepSeek V3 Base | DeepSeek | 89.3% | Lab | Dec 25 2024 |
| 4 | Claude 2 | Anthropic | 88% | Lab | Jul 11 2023 |
| 5 | Claude Instant 1.2 | Anthropic | 86.7% | Lab | Aug 9 2023 |


---

Canonical page: https://aireleasetracker.com/benchmark/gsm8k
Site index: https://aireleasetracker.com/llms.txt
Source: AI Release Tracker (https://aireleasetracker.com). Most benchmark scores come from lab launch material; gathered results identify the leaderboard that published them.