GPT-4.1

+ contribute

GPT-4.1 is a closed-weights language model from OpenAI, released April 2025.

62.3±0.5#173 of 719 LLMsfrom 69 benchmark results across 23 sources

$ how it compares

The LLMs ranked either side of GPT-4.1, out of 828 with a score, then more from OpenAI.

#	model	score	where	RAM	released
201	MiniMax-M1 (80k) MiniMax	62±1	api $0.96/M +local	~959GB	2025-06
202	Gemma 4 12B Google	62±1	local	~8.5GB	2026-06
203	GPT-4.1 OpenAI	62±1	api $3.50/M		2025-04
204	Nemotron 3.5 Lightning 30B-A3B NVIDIA	62±2	api $0.087/M +local	~20GB	2026-08
205	GLM-4.6V Z.ai	62±2	api $0.45/M +local	~51GB	2025-12
	more from OpenAI
1	GPT-6 Astra OpenAI	87±1	api $20.00/M		2026-09
3	GPT-6.1 Sol OpenAI	85±1	api $4.00/M		2026-09
7	GPT-6 Sol OpenAI	83±1	api $4.00/M		2026-09

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #154 of 300 · 26% of its score · 11 tests

Beats o1 (2024-12-17), o3-mini, DeepSeek R1 0528 here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, Jamba 1.5 Large here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
τ²-bench Telecom lab +4	34.0% ✎	▃·▅▅▃▃··█▃▃▅ beats 1/13	very hard · avg 10%	▇▇▇▇ 11%	sharp · at its level · several runs
SWE-bench Verified +1	48.5% ✎	▁··▃·▁▂▁▄█▄▁ beats 1/29	hard · avg 19%	▇▇▇ 7.0%	sharp · at its level · several runs
τ²-bench Retail +1	74.1% ✎	▅···█·▃····▃ beats 5/6	medium · avg 41%	▇ 2.5%	at its level · several runs
τ²-bench Airline +1	56.0% ✎	▄·▄··▄···█▄▄ beats 4/6	hard · avg 29%	▇ 1.6%	at its level · several runs
Finance Agent Benchmark v1 (537 questions, class-balanced accuracy)	26.7% ✎	█·▃▃·█▅▅▅▅▃▅ beats 14/18	very hard · avg 5.8%	▇ 1.4%	sharp
BFCL V4 (December 2025 archive)	54.0% ✎	▁▁▁▂██▅▅▆▂▂▂ beats 62/77	hard · avg 32%	▇ 1.0%	at its level
Terminal-Bench (pre-2.0 release)	28.4% ✎	▄·▄···▄▄···█ beats 3/5	very hard · avg 12%	▇ 0.5%	at its level
Galileo Agent Leaderboard v2 (action completion)	62.0% ✎	▅▃·▃▃▅▅█▃█▅▅ beats 19/19	hard · avg 27%	▇ 0.4%	at its level

3 more tests

test	result	vs the others in its run	difficulty	counts	why
BFCL v3	68.9% ✎	▁▁▁▁▁▃▅█▅▄▁▁ beats 107/114	medium · avg 52%	▇ 0.4%	blunt · at its level
KlinikeBench strict pass@1 (333 interactive clinical encounters, Terminus-2)	7.5% ✎	▁▁·▁▁▁▁▅▃█▅█ beats 6/27	very hard · avg 2.4%	▇ 0.3%	too hard for it
Galileo Agent Leaderboard v1 (tool selection quality)	91.8% ✎	▃▃▃▃▆▅▆▅█▃▅▂ beats 27/31	easy · avg 79%	▇ 0.2%	blunt

coding▁▁▁▃▆██▇▆▄▃▁ #142 of 458 · 19% of its score · 7 tests

Beats Gemini 2.5 Pro, Gemini 3.1 Flash-Lite, Qwen3-Coder-Next (80B-A3B) here, though they rank higher overall. Trails Devstral Small 2 24B, Llama 3.1 Nemotron Ultra 253B v1, Nemotron 3.5 Lightning 30B-A3B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
SWE-bench Verified +1	48.5% ✎	▁··▃·▁▂▁▄█▄▁ beats 1/29	hard · avg 19%	▇▇▇ 7.0%	sharp · at its level · several runs
Aider polyglot (225 exercises, two attempts)	52.4% ✎	▂·▂▃▅▂▃█▆▄▂▂ beats 28/49	very hard · avg 8.5%	▇▇ 6.1%	sharp · at its level
LiveCodeBench (version unspecified)	39.2% ✎	█·······▄▄▄▄ beats 0/5	hard · avg 21%	▇ 1.7%	at its level
HumanEval+	93.3% ✎	▄▄▄·····▄▄·█ beats 4/6	easy · avg 67%	▇ 1.3%	sharp
MBPP+	79.4% ✎	▃·····▃█▃··▃ beats 1/6	medium · avg 61%	▇ 1.2%	typical
WebApp1K-Duo-React pass@1 (onekq-ai)	68.8% ✎	▁··▁·▁▁▁▆█▇█ beats 29/45	hard · avg 32%	▇ 1.0%	at its level
HumanEval (simple-evals) lab	94.5% ✎	▆█▆▆▃··▂···▂ beats 13/15	easy · avg 84%	▇ 0.5%	maxed out

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #99 of 393 · 6.8% of its score · 4 tests

Beats LongCat 2.0 (1.6T-A48B), Inkling Small (266B-A12B), GPT-5.4 nano here, though they rank higher overall. Trails Mistral Large 3 (675B-A41B) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Arena-Hard v2.0 hard prompts (style control, Gemini-2.5 judge)	50.0% ✎	▂·▃▅▅·▃█▆▂▂▂ beats 15/22	very hard · avg 12%	▇ 3.2%	sharp · at its level
LMArena (text)	1414.8 ✎	▁▁▁▂▃▃▃▆▆▇█▃ beats 263/379	avg 1268	▇ 2.0%	blunt
IFEval (aggregation unspecified)	85.6% ✎	█··█·█··████ beats 1/6	easy · avg 75%	▇ 0.9%	typical
Arena-Hard v2.0 creative writing (GPT-4.1 + Gemini-2.5 judges)	61.5% ✎	▆▂▂▆▂▂▃█▂▃·▂ beats 17/22	hard · avg 19%	▇ 0.6%	sharp · at its level · noisy

math▁▂▃▅▆▅▄█▄▁▁▁ #182 of 398 · 14% of its score · 9 tests

Beats MiniMax-M2.1 here, though they rank higher overall. Trails Phi-4-mini-reasoning 3.8B, OpenThinker3 7B, Olmo 3 7B Think here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
FrontierMath tiers 1–3 v2	6.0% ✎	▂··▃▁▃▄▄█▅▃▃ beats 4/82	very hard · avg 1.2%	▇▇ 4.4%	sharp
MATH level 5	83.0% ✎	▂▂▅▆▄█▆▃▅▂▃▄ beats 69/92	hard · avg 32%	▇ 2.7%	sharp
OTIS mock AIME 2024–2025	38.3% ✎	▃▃▄▃▃▃▇▇█▃▂▅ beats 71/193	very hard · avg 8.5%	▇ 2.1%	sharp · at its level · noisy
MATH-500 (simple-evals) lab	82.1% ✎	▃··▅▅▃···▃·█ beats 2/9	medium · avg 61%	▇ 1.5%	sharp
AIME 2025	32.0% ✎	▄▄···▄▄··▄·█ beats 0/6	hard · avg 19%	▇ 1.0%	sharp · at its level · noisy
MathNet-Solve-Test micro-average accuracy (6,400 olympiad problems)	21.4% ✎	▃▃···█▃·▃·▃▃ beats 1/8	hard · avg 15%	▇ 0.7%	at its level
AIME 2024	47.0% ✎	▁▁▁▂▁▃▁▅███▃ beats 21/85	hard · avg 32%	▇ 0.6%	sharp · at its level · noisy
MATH-500	90.6% ✎	▄·▄··█··▄·▄▄ beats 0/6	very easy · avg 86%	▇ 0.6%	maxed out · noisy

1 more test

test	result	vs the others in its run	difficulty	counts	why
MGSM (simple-evals) lab	86.9% ✎	▁····▁▅▁█▁▃▁ beats 3/16	easy · avg 82%	▇ 0.5%	blunt

vision & documents▁▁▃▂▁▃▄▂▆▇█▄ #72 of 147 · 4.1% of its score · 3 tests

Beats DeepSeek V4 Flash Vision (Exp), GPT-5.4 nano, GPT-5 mini here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
Arena vision (style controlled)	1214.0 ✎	▁▁▂▂▂▂▄▅▅▆██ beats 78/145	avg 1068	▇ 1.4%	blunt
CharXiv (reasoning questions) lab	56.7% ✎	▄··▂█··▂···▄ beats 5/9	hard · avg 34%	▇ 1.4%	at its level
CharXiv descriptive questions lab	87.9% ✎	▄▄·····▄·▄█▄ beats 3/6	easy · avg 72%	▇ 1.3%	typical

web search▂▁▂▆▄█▃▃▁▃·▁ #3 of 80 * · 1.4% of its score · 1 test

Beats GPT-6 Astra, Claude Opus 5.5, Claude Fable 5.1 here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
Finance Agent Benchmark v1 (537 questions, class-balanced accuracy)	26.7% ✎	█·▃▃·█▅▅▅▅▃▅ beats 14/18	very hard · avg 5.8%	▇ 1.4%	sharp

other tests28 not in a task · 37% of its score

test	result	vs the others in its run	difficulty	counts	why
GPQA Diamond +1	66.9% ✎	▁▁▁▁▂▄▆▄▅▇█▅ beats 100/208	hard · avg 40%	▇▇ 6.0%	sharp · at its level · several runs
GPQA Diamond (simple-evals) lab	66.3% ✎	█▃▃█·▃▃▅▃▃▃▅ beats 10/16	hard · avg 40%	▇▇ 4.7%	sharp · at its level
MMLU	89.6% ✎	▄·····█▄▄▄·▄ beats 1/6	medium · avg 71%	▇ 2.6%	typical
KMMLU-HARD (Daekeun Kim, zero-shot)	42.8% ✎	█···▁▁▁▃▃·▆▂ beats 11/23	very hard · avg 29%	▇ 2.2%	sharp
CLIcK Korean (Daekeun Kim)	80.5% ✎	▆▄▄▄█▄▂·▄█·▄ beats 14/23	medium · avg 58%	▇ 2.1%	typical
MMLU-Pro	81.7% ✎	█▄·▄···▄··▄▄ beats 0/6	medium · avg 59%	▇ 2.0%	typical
HAE-RAE 1.0 (Daekeun Kim)	85.8% ✎	▄▄▄▂▂█▂▄▂█▂▄ beats 14/22	medium · avg 63%	▇ 1.7%	typical
ARC-AGI-1 public evaluation set (ARC Prize model_baseline, pass@2)	11.8% ✎	▁·▇▃▁·█▇▁·▇▃ beats 5/27	very hard · avg 4.1%	▇ 1.5%	sharp · at its level · noisy

20 more tests

test	result	vs the others in its run	difficulty	counts	why
KCL-Essay vanilla	42.4% ✎	▁···▁▃▂█▂▃▃▂ beats 16/23	hard · avg 18%	▇ 1.5%	at its level
EQ-Bench 3 (emotional intelligence, judge Elo)	1206.0 ✎	▁·▂▁▄▅▅▃█▄▁▁ beats 31/62	avg 1044	▇ 1.3%	typical
CMMLU	77.6% ✎	█····▄··▄▄▄▄ beats 0/6	medium · avg 67%	▇ 1.2%	typical
C-Eval	79.5% ✎	█····▄·▄·▄▄▄ beats 1/6	medium · avg 68%	▇ 1.2%	typical
MMLU (simple-evals, zero-shot CoT) lab	90.2% ✎	▂▂·█▄▄▂·█▂·▂ beats 11/16	easy · avg 80%	▇ 1.1%	typical
AraGen-v3 3C3H score (Inception, Claude Sonnet 4 judge)	74.5% ✎	▁▁·▁▂▂▄█▄▅▄▄ beats 58/66	hard · avg 31%	▇ 1.1%	at its level
DROP	66.9% ✎	▄···▄·▄█·▄·▄ beats 1/6	medium · avg 41%	▇ 0.9%	at its level · noisy
HindiGen-v1 3C3H score (Inception, Claude 3.5 Sonnet judge)	73.4% ✎	▁▁▁▁·▂·█▆█▇▁ beats 23/31	medium · avg 48%	▇ 0.9%	at its level
COLLIE (constrained text generation)	50.0% ✎	▄···▄█▄····█ beats 3/6	hard · avg 30%	▇ 0.8%	at its level
VitaBench	19.0% ✎	█·······▄█·█ beats 3/6	very hard · avg 6.4%	▇ 0.8%	typical
Open Persian LLM Leaderboard v2 average (21 tasks)	74.4% ✎	▂▄▂▃█▄▃▆▃▅▃▂ beats 78/81	medium · avg 58%	▇ 0.7%	blunt · at its level
KCL-MCQA vanilla	47.9% ✎	▂···█▂▆▆██▆▂ beats 16/23	very hard · avg 30%	▇ 0.7%	at its level
SimpleQA Verified	31.1% ✎	▁▁▃▂▃▃▆▆█▃▂▅ beats 21/78	very hard · avg 8.0%	▇ 0.6%	typical
Chess Puzzles (Epoch)	6.0% ✎	█▃▁▁▁▂▁▄▂▂▁▁ beats 49/135	very hard · avg 5.9%	▇ 0.4%	sharp · too hard for it · noisy
ARC-AGI-2 public evaluation set (ARC Prize model_baseline, pass@2)	0.0% ✎	█··▂▅▁▃▁▃·▁▃ beats 2/24	very hard · avg 0.3%	▇ 0.4%	sharp · too hard for it · noisy
BeyondAIME	22.1% ✎	█····█·▄▄··▄ beats 1/6	hard · avg 25%	▇ 0.3%	at its level
SimpleQA (simple-evals) lab	41.6% ✎	▄·▁▁▁▁··█▁·▁ beats 10/14	hard · avg 18%	▇ 0.2%	at its level · noisy
DROP F1, 3-shot (simple-evals) lab	79.4% ✎	▂▂▄▆▆█·▂···▄ beats 2/16	easy · avg 78%	▇ 0.2%	blunt
Vectara hallucination leaderboard, 2025 dataset (HHEM-2.3, factual consistency rate)	98.0% ✎	▁▁▂▃▄▄█▆▃▂▂▁ beats 121/140	very easy · avg 94%	▇ 0.1%	blunt · maxed out · noisy
Vectara hallucination leaderboard (HHEM-2.3, factual consistency rate)	94.4% ✎	▁▂▁▂▅█▂▃▄▃▁▁ beats 80/98	very easy · avg 90%	▇ 0.0%	blunt

$ where to run it

Through an API

api · $3.50/M · $2 in / $8 out per 1M tokens · 1M context · platform.openai.com ↗ · price sourceStandard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ.

$ how the score was worked out · Meituan LongCat-Flash-Chat mo… 25%, Epoch AI — internal evaluatio… 19%, OpenAI simple-evals README be… 9% and 20 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The 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's added up by who ran the tests. How it works

source	results	share of the score	licence
Meituan LongCat-Flash-Chat model card (evaluation table)	21	▇▇▇▇▇▇▇ 25%	MIT
Epoch AI — internal evaluations	7	▇▇▇▇▇▇ 19%	CC BY 4.0
OpenAI simple-evals README benchmark results (OpenAI's own models) · lab-reported	7	▇▇▇ 9%	MIT
Aider polyglot benchmark runs	1	▇▇ 6%	Apache-2.0
Daekeun Kim Korean LLM benchmark runs	3	▇▇ 6%	MIT
MUA-RL: Multi-turn User-interacting Agent Reinforcement Learning for agentic tool use (arXiv:2508.18669v1)	1	▇▇ 5%	CC-BY-4.0
Arena-Hard-Auto v2.0-Preview leaderboard (repository README)	2	▇ 4%	Apache-2.0
OpenAI vendor-reported benchmark scores for its own models (openai.com announcement posts and Deployment Safety Hub system cards) · lab-reported	3	▇ 4%	Factual vendor-reported score (not licensed content)
τ²/τ³-bench leaderboard: Sierra's own evaluation runs	3	▇ 4%	MIT
LMArena leaderboard dataset	2	▇ 3%	CC BY 4.0
Korean Canonical Legal Benchmark: Toward Knowledge-Independent Evaluation of LLMs’ Legal Reasoning Capabilities	2	▇ 2%	CC BY 4.0
Moonshot AI Kimi-K2-Instruct model card (instruction-model evaluation table)	1	▇ 2%	Modified MIT License (Kimi K2)
ARC Prize Foundation model_baseline outputs on ARC-AGI-1/2 public evaluation sets	2	▇ 2%	MIT
Inception Arabic/Hindi leaderboards (AraGen-v3, HindiGen-v1, Arabic IFEval)	2	▇ 2%	Apache-2.0
Berkeley Function Calling Leaderboard result archive	2	▇ 1%	Apache-2.0
Finance Agent Benchmark (Vals AI, arXiv:2508.00828v1)	1	▇ 1%	CC-BY-4.0
EQ-Bench 3 canonical leaderboard results	1	▇ 1%	MIT
WebApp1K models leaderboard (onekq-ai)	1	▇ 1%	MIT
Open Persian LLM Leaderboard v2 (Part DP AI and AUT NLP Lab)	1	▇ 1%	Apache-2.0
MathNet: a Global Multimodal Benchmark for Mathematical Reasoning and Retrieval — authors' own model runs	1	▇ 1%	CC BY 4.0
Galileo Agent Leaderboard (v1 and v2 results)	2	▇ 1%	Apache-2.0
KlinikeBench: Evaluating Language Models Beyond Diagnostic Accuracy (arXiv:2609.38480v2)	1	▇ 0%	CC BY 4.0
Vectara hallucination leaderboard (vectara/hallucination-leaderboard repository)	2	▇ 0%	Apache-2.0
prior · pull toward the middle		▇ 0%
weighted average		= 62.3