[whatmodel.sh](../../README.md) / GPT-4.5 Preview # GPT-4.5 Preview OpenAI · released 2025-02 · proprietary **score 62.6** ±0.8 · comparison only: no current run option verified · from 23 benchmark results across 12 sources Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only. ## What it's good at ● marks GPT-4.5 Preview among every model on that task or test, worst to best. 
task                vs the others          rank  counts  tests
agent tasks         ▁▁▂▆█●▅▄▄▃▁▁   #187 of 354*    1.9%      2
coding              ▁▁▁▃▄▆█●▅▄▃▁    #183 of 539     21%      3
chat & writing      ▁▁▁▃▅▄▇█▇▇●▃     #77 of 474     15%      3
math                ▁▁▃▆██●█▃▁▁▁    #215 of 530     17%      4
vision & documents  ▁▁▂▃▂▃▅▄●██▄     #81 of 189     12%      3
agent tasks · #187 of 354 · 1.9% of its score Beats [o1 (2024-12-17)](o1-2024-12-17.md), [o3-mini](o3-mini.md), [DeepSeek R1 0528](deepseek-r1-0528.md) here, though they rank higher overall. Trails [SmolLM3 3B](smollm3-3b.md), [Jamba 1.5 Mini](jamba-1-5-mini.md), [Jamba 1.5 Large](jamba-1-5-large.md) here, though they rank lower overall. 
test                                result  vs the others in its run    difficulty        counts  why
BFCL v3                              70.3%  ▁▁▁▁▁▃▅█▅▄●▁ beats 109/114  medium · avg 51%    1.2%  blunt · at its level
Galileo Agent Leaderboard v1 (too…   90.0%  ▃▃▃▃▆▅▆▅●▂▅▂ beats 22/31    easy · avg 78%      0.7%  typical
coding · #183 of 539 · 21% of its score Beats [Qwen3-Coder-Next (80B-A3B)](qwen3-coder-next.md), [Gemini 2.5 Pro](gemini-2-5-pro.md), [Gemini 3.1 Flash-Lite](gemini-3-1-flash-lite.md) here, though they rank higher overall. Trails [QwQ-32B-Preview](qwq-32b-preview.md), [Ministral 3 8B Reasoning 2512](ministral-3-8b-reasoning.md), [Phi-4-reasoning 14B](phi-4-reasoning.md) here, though they rank lower overall. 
test                                result  vs the others in its run  difficulty            counts  why
Aider polyglot (225 exercises, tw…   44.9%  ▂·▂▃▅▂▃●▆▄▂▂ beats 22/49  very hard · avg 7.6%     16%  sharp · at its level
WebApp1K-Duo-React pass@1 (onekq-…   66.3%  ▁··▁·▁▁▁▆●▇█ beats 22/45  hard · avg 29%          3.5%  at its level
HumanEval (simple-evals) (maker)     88.6%  ▆●▆▆▂··▂···▂ beats 6/15   easy · avg 83%          1.3%  maxed out
chat & writing · #77 of 474 · 15% of its score Beats [Claude Haiku 5.5](claude-haiku-5-5.md), [DeepSeek V4 Flash (0731)](deepseek-v4-flash.md), [GPT-5.2](gpt-5-2.md) here, though they rank higher overall. 
test                                result  vs the others in its run    difficulty           counts  why
Arena-Hard v2.0 hard prompts (sty…   50.0%  ▂·▃▅▅·▃●▆▂▂▂ beats 14/22    very hard · avg 12%    7.3%  sharp · at its level
LMArena (text)                      1444.6  ▁▁▁▂▃▃▃▆▆▇●▃ beats 311/385  avg 1260               6.0%  blunt
Arena-Hard v2.0 creative writing …   51.4%  ▆▂▂▆▂▂●█▂▃·▂ beats 12/22    hard · avg 19%         1.5%  sharp · at its level · noisy
math · #215 of 530 · 17% of its score Beats [MiniMax-M2.1](minimax-m2-1.md) here, though they rank higher overall. Trails [Mixtral 8x7B v0.1 (base)](mixtral-8x7b-v0-1.md), [Phi-4-mini-reasoning 3.8B](phi-4-mini-reasoning.md), [OpenThinker3 7B](openthinker3-7b.md) here, though they rank lower overall. 
test                             result  vs the others in its run   difficulty            counts  why
MATH level 5                      78.6%  ▂▂▅▆▄█▆●▅▂▃▄ beats 65/92   hard · avg 28%          7.0%  sharp
OTIS mock AIME 2024–2025          37.8%  ▃▃▄▃▃●▇▇█▃▂▅ beats 69/197  very hard · avg 7.2%    4.8%  sharp · at its level · noisy
MATH-500 (simple-evals) (maker)   87.1%  ▃··▅●▃···▃·█ beats 4/9     medium · avg 61%        3.4%  sharp
MGSM (simple-evals) (maker)       86.9%  ▁····▁●▁█▁▃▁ beats 2/16    easy · avg 80%          1.7%  typical
vision & documents · #81 of 189 · 12% of its score Beats [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-5](gpt-5.md), [o3](o3.md) here, though they rank higher overall. 
test                                result  vs the others in its run   difficulty      counts  why
Arena vision (style controlled)     1225.0  ▁▁▂▂▂▂▃▄▅●█▇ beats 86/152  avg 1061          4.0%  blunt
CharXiv (reasoning questions) (ma…   55.4%  ▄··▂●··▂···▄ beats 4/9     hard · avg 32%    3.9%  at its level
CharXiv descriptive questions (ma…   90.0%  ▄▄·····▄·▄█● beats 6/6     easy · avg 71%    3.7%  typical
other tests (8) 
test                                result  vs the others in its run    difficulty            counts  why
GPQA Diamond (simple-evals) (make…   69.5%  █▃▃█·▃▃●▃▃▃▅ beats 11/16    hard · avg 39%           11%  sharp · at its level
GPQA Diamond                         68.7%  ▁▁▁▁▂▄▆▄●▇█▅ beats 103/211  hard · avg 39%           10%  sharp · at its level
ARC-AGI-1 public evaluation set (…   15.6%  ▁·▇●▁·█▇▁·▇▃ beats 6/27     very hard · avg 4.0%    4.3%  sharp · at its level · noisy
EQ-Bench 3 (emotional intelligenc…  1153.7  ▁·▂▁▄●▅▃█▄▁▁ beats 21/62    avg 1034                3.7%  typical
MMLU (simple-evals, zero-shot CoT…   90.8%  ▂▂·█▄▄▂·●▂·▂ beats 13/16    easy · avg 79%          3.4%  typical
SimpleQA (simple-evals) (maker)      62.5%  ▄·▁▁▁▁··█▁·● beats 14/14    hard · avg 17%          0.6%  at its level · noisy
DROP F1, 3-shot (simple-evals) (m…   83.4%  ▂▂▄▆▆●·▂···▄ beats 11/16    easy · avg 78%          0.5%  blunt
Vectara hallucination leaderboard…   98.8%  ▁▁▂▃▄▄█▆▃▂●▁ beats 134/140  very easy · avg 94%     0.2%  blunt · maxed out · noisy
## How the score was worked out Each result implies a score on its own, after adjusting for how hard that test is and who ran it. The model's score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its *counts* column above; here it is added up by who ran the tests. [Method](../method.md) 
source                                    results  share of the score
OpenAI simple-evals README benchmark re…        7   22% ▇▇▇
Epoch AI                                        3   22% ▇▇▇
Aider polyglot benchmark runs                   1   16% ▇▇
LMArena leaderboard dataset                     2   10% ▇
Arena-Hard-Auto v2.0-Preview leaderboard        2    9% ▇
OpenAI vendor-reported benchmark scores…        2    8% ▇
ARC Prize Foundation model_baseline out…        1    4% ▇
EQ-Bench 3 canonical leaderboard results        1    4% ▇
WebApp1K models leaderboard                     1    3% ▇
Berkeley Function Calling Leaderboard r…        1    1% ▇
Galileo Agent Leaderboard                       1    1% ▇
Vectara hallucination leaderboard               1    0% ▇
prior: pull toward the middle                        0% ▇
weighted average                                   = 62.6
Licences: OpenAI simple-evals README benchmark results ([MIT](../../licenses/openai-simple-evals-MIT.txt)) · Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Aider polyglot benchmark runs ([Apache-2.0](../../licenses/aider-polyglot-runs-Apache-2.0.txt)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Arena-Hard-Auto v2.0-Preview leaderboard ([Apache-2.0](../../licenses/arena-hard-auto-Apache-2.0.txt)) · ARC Prize Foundation model\_baseline outputs on ARC-AGI-1/2 public evaluation sets ([MIT](../../licenses/arcprize-model-baseline-outputs-MIT.txt)) · EQ-Bench 3 canonical leaderboard results ([MIT](../../licenses/eqbench3-MIT.txt)) · WebApp1K models leaderboard ([MIT](../../licenses/webapp1k-leaderboard-MIT.txt)) · Berkeley Function Calling Leaderboard result archive ([Apache-2.0](../../licenses/bfcl-result-archive-v4-Apache-2.0.txt)) · Galileo Agent Leaderboard ([Apache-2.0](../../licenses/galileo-agent-leaderboard-Apache-2.0.txt)) · Vectara hallucination leaderboard ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) ## How to run it No verified API or local download yet. [interactive version](https://whatmodel.sh/model/gpt-4-5-preview-2025-02-27) · [JSON](https://whatmodel.sh/api?format=json&all=1)