GPT-4.1 mini

GPT-4.1 mini is a non-reasoning model from OpenAI in the GPT-4.1 family. 28 benchmarks count toward its score, in 7 categories.

availableShows if the model has enough results for an index.
IndexOverall score out of 100.34.8 ±3.6
CoverageShare of the index weight with results.95%
SpeedOutput tokens per second.38/s
Input / 1MUS dollars per 1M input tokens.$0.4 batch $0.2
Output / 1MUS dollars per 1M output tokens.$1.6 batch $0.8 US dollars per 1M output tokens in a batch.
ContextMaximum tokens in one request.1.05M
EloLMArena rating and rank.1340 (#199)

The index is a score out of 100. The ± range shows how much it can change. Batch work costs less.

38,631 votes. Elo shows what people prefer. It does not change the score.

CapabilitiesScore per category, out of 100.

Out of 100
AgenticMulti-step tasks with tools.
35.5
CodingCode writing and repair.
29.0
ReasoningLogic problems and puzzles.
36.6
MultimodalTasks with images and text.
40.5
KnowledgeFacts and expert knowledge.
36.5
MultilingualTasks in many languages.
N/A
InstructionTasks with strict rules in the prompt.
35.0
MathMath problems.
37.5

Results

28 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result as a score out of 100.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
Instruction-Following EvalInstruction88.5%42.0Jeffrey Zhou et al.
MATH 500Math88.0%40.29 Jan 2026Vals AI
MGSMMultilingual87.8%9 Jan 2026Vals AI
Massive Multitask Language UnderstandingKnowledge87.5%Dan Hendrycks et al.
MATH level 5Math87.3%41.6Epoch AI
MMLU ProKnowledge77.2%42.21 Sept 2026Vals AI
MMMU ProMultimodal70.5%42.11 Sept 2026Vals AI
GPQA DiamondKnowledge67.9%41.01 Sept 2026Vals AI
Artificial Analysis GPQA DiamondKnowledge66.4%36.6Artificial Analysis
GPQA diamondKnowledge65.8%39.0Epoch AI
Graduate-Level Google-Proof Q&AKnowledge64.2%37.5David Rein et al.
τ²-bench RetailAgentic61.4%42.9Sierra30 Apr 2026Sierra Research
Artificial Analysis MMMU-ProMultimodal58.7%38.8Artificial Analysis
LiveCodeBenchCoding58.2%37.21 Sept 2026Vals AI
τ²-Bench Tool-Agent-User EvaluationAgentic52.9%36.8Victor Barres et al.
AIMEMath49.4%36.716 Apr 2026Vals AI
τ²-bench TelecomAgentic48.9%33.9Sierra2 Mar 2026Sierra Research
τ²-bench AirlineAgentic48.7%33.8Sierra2 Mar 2026Sierra Research
OTIS Mock AIME 2024-2025Math44.7%36.6Epoch AI
Artificial Analysis Long Context ReasoningReasoning44.0%38.7Artificial Analysis
Artificial Analysis IFBenchInstruction38.3%28.0Artificial Analysis
SWE-bench VerifiedCoding23.9%16.7mini-SWE-agent26 Feb 2026SWE-bench team
SWE-bench VerifiedCoding23.9%16.7mini-SWE-agent1 Sept 2026SWE-bench team
Software Engineering Benchmark VerifiedCoding23.6%16.4Carlos E. Jimenez et al.
Artificial Analysis Omniscience AccuracyKnowledge20.3%39.0Artificial Analysis
Artificial Analysis Coding IndexCoding20.2%33.2Artificial Analysis
SimpleQA VerifiedKnowledge12.7%33.3Epoch AI
Artificial Analysis Intelligence IndexKnowledge10.2%35.1Artificial Analysis
Mystery Game PuzzlesReasoning7.0%41.4Epoch AI
Chess PuzzlesReasoning7.0%32.6Epoch AI
FrontierMath-Tiers-1-3-v2-PrivateMath6.7%35.2Epoch AI
Artificial Analysis Humanity's Last ExamKnowledge5.0%30.0Artificial Analysis
FrontierMath-2025-02-28-PrivateMath4.5%34.8Epoch AI
ARC-AGI-1 (semi-private)Reasoning3.5%28.9ARC Prize Foundation
GDPval-AA normalizedAgentic0.0%35.6Artificial Analysis
Critical Physics TasksReasoning0.0%38.4Artificial Analysis
ARC-AGI-2 (semi-private)Reasoning0.0%39.7ARC Prize Foundation

28 benchmarks count, from 35 of 37 results. A grey row does not count. Too few models took that benchmark.

Sources

BenchLM benchmark aggregationUsed with attribution; per-benchmark results credited to their original authorsOpenRouter, collected directlyNo licence statedVals AI, collected directlyNo licence stated. Read from the public leaderboard and credited to Vals AIEpoch AI, collected directlyCC BY — free to use and redistribute with attributionSierra Research, collected directlyMIT — results are in the licensed repositorySWE-bench team, collected directlyNo licence stated for the leaderboard. The harness repo is MITARC Prize Foundation, collected directlyNo licence stated. Their terms ask for written permission before commercial use

Same level, lower price

Qwen3.8-Flash-Next66.4 · Freedots3-note Preview65.5 · FreeApodex 1.1 Mini59.1 · Free

More from OpenAI

GPT-6 Astra83.5GPT-5.6 Sol76.7GPT-5.6 Terra73.4GPT-6 Sol76.9GPT-5.572.3