MathVista

Name: MathVista
Creator: MathVista authors and website maintainers
License: https://creativecommons.org/licenses/by-sa/4.0/

We show this table for reference; we do not rank on it.

Data verified June 9, 2025

MathVista tests mathematical reasoning about images. This table shows published accuracy on the testmini split, with each source configuration kept separate. The results are display only and do not enter overall or category rankings.

MathVista by MathVista authors and website maintainers and contributors. Original benchmark · CC BY-SA 4.0. Selected aggregate scores and adjacent metadata; converted numeric strings to numbers and missing-score markers to null; separated testmini from test; added configuration notes and three named profile references. Selected aggregate-result compilation and its adaptation. Task content, source images, model outputs, and model weights are excluded.

Accuracy (%) on MathVista — June 9, 2025

We mirror the published accuracy (%) view for MathVista. DreamPRM (o4-mini) has the highest reported value at 85.20%. The compilation contains different evaluation setups; the highest reported value does not establish a comparable model ranking. We do not use these results to rank models overall.

DreamPRM (o4-mini)

DreamPRM selection pipeline with o4-mini; separate from a bare o4-mini evaluation; Source date 2025-06-04 (compiler field)

85.20%

VL-Rethinker

Published method: Reason 🧠; Additional checkpoint and evaluation settings unverified; Source date 2025-04-10 (compiler field)

80.30%

Step R1-V-Mini

Published method: Reason 🧠; Additional checkpoint and evaluation settings unverified; Source date 2025-04-07 (compiler field)

80.10%

78 model or configuration recordsMultimodal & GroundedHistorical source configurationsDisplay onlyUpdated June 9, 2025

MathVista testmini: 78 published model or configuration records
Source configuration	Accuracy	Evaluation setup and sources	Parameters (B)	Open / closed
DreamPRM (o4-mini)	85.2%	DreamPRM selection pipeline with o4-mini; separate from a bare o4-mini evaluation; Source date 2025-06-04 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 87.7% geometry problem solving 95.7% math word problem 83.9% textbook question answering 82.3% visual question answering 73.2% algebraic reasoning 94.1% arithmetic reasoning 80.8% geometry reasoning 95.5% logical reasoning 75.7% numeric commonsense 64.5% scientific reasoning 81.4% statistical reasoning 91.1%	Not reported	Unknown
VL-Rethinker	80.3%	Published method: Reason 🧠; Additional checkpoint and evaluation settings unverified; Source date 2025-04-10 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Step R1-V-Mini	80.1%	Published method: Reason 🧠; Additional checkpoint and evaluation settings unverified; Source date 2025-04-07 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 80.3% geometry problem solving 89.9% math word problem 87.6% textbook question answering 77.2% visual question answering 63.1% algebraic reasoning 86.1% arithmetic reasoning 75.6% geometry reasoning 87.4% logical reasoning 54.1% numeric commonsense 56.3% scientific reasoning 77.9% statistical reasoning 87.7%	Not reported	Unknown
Kimi-k1.6-preview-20250308	80%	Published method: Reason 🧠; Additional checkpoint and evaluation settings unverified; Source date 2025-03-10 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 82.9% geometry problem solving 91.8% math word problem 82.3% textbook question answering 76% visual question answering 63.1% algebraic reasoning 87.2% arithmetic reasoning 73.1% geometry reasoning 89.1% logical reasoning 45.9% numeric commonsense 52.8% scientific reasoning 77% statistical reasoning 90.7%	Not reported	Unknown
Doubao-pro-1.5	79.5%	Published method: Reason 🧠; Additional checkpoint and evaluation settings unverified; Source date 2025-01-22 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 77.7% geometry problem solving 88.94% math word problem 86.02% textbook question answering 82.28% visual question answering 62.01% algebraic reasoning 89.67% arithmetic reasoning 73.45% geometry reasoning 88.29% logical reasoning 35.14% numeric commonsense 55.32% scientific reasoning 77.12% statistical reasoning 86.47%	Not reported	Unknown
Ovis2_34B	77.1%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2025-02-10 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 75.46% geometry problem solving 84.62% math word problem 89.25% textbook question answering 72.78% visual question answering 62.01% algebraic reasoning 80.07% arithmetic reasoning 72.24% geometry reasoning 82.01% logical reasoning 37.84% numeric commonsense 54.86% scientific reasoning 73.77% statistical reasoning 85.05%	Not reported	Unknown
Kimi-k1.5	74.9%	Published method: Reason 🧠; Additional checkpoint and evaluation settings unverified; Source date 2025-01-22 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
OpenAI o1The original source reports research o1, not o1-preview or released o1	73.9%	Research o1; maximal test-time compute; vision enabled; exact checkpoint undisclosed; Source date 2024-09-12 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Llama 4 MaverickCheckpoint identity and catalog date require review; source configuration retained	73.7%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2025-04-05 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Vision-R1-7B	73.2%	Published method: Reason 🧠; Additional checkpoint and evaluation settings unverified; Source date 2025-03-09 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 75.1% geometry problem solving 82.7% math word problem 80.1% textbook question answering 70.3% visual question answering 54.7% algebraic reasoning 78.6% arithmetic reasoning 65.2% geometry reasoning 79.1% logical reasoning 40.5% numeric commonsense 45.1% scientific reasoning 68.9% statistical reasoning 83.1%	Not reported	Unknown
Gemini 2.0 Flash	73.1%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2025-02-05 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
QVQ-72B-Preview	71.4%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-12-24 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Qwen2VL-72BSource metadata links a QVQ card; sibling checkpoint settings were not inherited	70.5%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-12-24 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Pixtral Large (124B)Source folder says Pixtral-Large 124B; score filename alone does not establish model identity	69.4%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-11-18 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Grok-2	69%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-08-13 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Grok-2 mini	68.1%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-08-13 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Claude 3.5 SonnetJune 2024 result; source metadata says June 20 and the provider page says June 21	67.7%	Initial June 2024 Sonnet; zero-shot chain-of-thought; exact API snapshot undisclosed; Source date 2024-06-20 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
LLaVA-OneVision	67.5%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-08-06 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
InternVL2-8B-MPO	67.3%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-11-14 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 72.49% geometry problem solving 73.56% math word problem 69.89% textbook question answering 66.46% visual question answering 50.28% algebraic reasoning 70.11% arithmetic reasoning 57.51% geometry reasoning 71.55% logical reasoning 27.03% numeric commonsense 43.06% scientific reasoning 65.57% statistical reasoning 79.07%	Not reported	Unknown
Ovis1.6-Gemma2-9BSource metadata links an Ovis2 card; sibling checkpoint settings were not inherited	67.2%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2025-02-10 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
InternVL2-Pro	66.8%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-09-04 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 70.63% geometry problem solving 65.38% math word problem 76.88% textbook question answering 71.52% visual question answering 48.04% algebraic reasoning 66.55% arithmetic reasoning 62.32% geometry reasoning 63.6% logical reasoning 27.03% numeric commonsense 40.28% scientific reasoning 65.57% statistical reasoning 81.06%	Not reported	Unknown
Chimera-Reasoner-8B	66.2%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-12-05 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 65.43% geometry problem solving 71.63% math word problem 74.73% textbook question answering 66.46% visual question answering 51.96% algebraic reasoning 68.33% arithmetic reasoning 58.92% geometry reasoning 70.29% logical reasoning 21.62% numeric commonsense 45.83% scientific reasoning 61.48% statistical reasoning 73.09%	Not reported	Unknown
TextGrad (GPT-4o)	66.1%	TextGrad pipeline with GPT-4o; separate from a bare GPT-4o evaluation; Source date 2024-07-08 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
TBAC-VLR1-3B-preview	64.8%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2025-04-16 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Gemini 1.5 Pro (May 2024)	63.9%	May 2024 source result; separate from the February 2024 configuration; Source date 2024-05-17 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
GPT-4o	63.8%	May 2024 source result; exact API snapshot and sampling settings unverified; Source date 2024-05-13 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Phi-4-multimodal-insNamed profile referenceSource date 2024-03-03 conflicts with the 2025 release; retained as a compiler-field error	62.4%	Microsoft internal platform; zero-shot vision prompts; original benchmark images; sampling settings undisclosed; Source date 2024-03-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Open
InternVL-Chat-V1.2-Plus	59.9%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-02-22 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 51.67% geometry problem solving 61.06% math word problem 79.57% textbook question answering 52.53% visual question answering 56.98% algebraic reasoning 54.45% arithmetic reasoning 63.17% geometry reasoning 61.09% logical reasoning 16.22% numeric commonsense 48.61% scientific reasoning 55.74% statistical reasoning 60.8%	Not reported	Unknown
Gemini 1.5 Flash (May 2024)	58.4%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-05-17 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
GPT-4T 2024-04-09	58.1%	April 9, 2024 source configuration; prompting and sampling settings unverified; Source date 2024-05-13 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Pixtral-12B	58%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-10-09 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
InternLM-XComposer2-VL-7B	57.6%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-01-22 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 55.02% geometry problem solving 62.98% math word problem 73.66% textbook question answering 56.33% visual question answering 39.66% algebraic reasoning 56.58% arithmetic reasoning 52.41% geometry reasoning 62.34% logical reasoning 8.11% numeric commonsense 42.36% scientific reasoning 59.02% statistical reasoning 64.12%	Not reported	Unknown
Gemini 1.0 Ultra	53%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-12-06 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 49.07% geometry problem solving 56.25% math word problem 53.76% textbook question answering 68.99% visual question answering 40.22% algebraic reasoning 58.36% arithmetic reasoning 45.89% geometry reasoning 55.65% logical reasoning 21.62% numeric commonsense 38.89% scientific reasoning 62.3% statistical reasoning 59.47%	Not reported	Unknown
Grok-1.5V	52.8%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-04-12 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Gemini 1.5 Pro (Feb 2024)	52.1%	February 2024 source result; separate from the May 2024 configuration; Source date 2024-02-15 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Claude 3 OpusNamed profile referenceNamed profile reference; the source configuration remains the evaluation identity	50.5%	Zero-shot chain-of-thought; exact API snapshot undisclosed; Source date 2024-03-04 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Closed
GPT-4V (Playground)	49.9%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-10-15 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 43.12% geometry problem solving 50.48% math word problem 57.53% textbook question answering 65.19% visual question answering 37.99% algebraic reasoning 53.02% arithmetic reasoning 49.01% geometry reasoning 51.05% logical reasoning 21.62% numeric commonsense 20.14% scientific reasoning 63.11% statistical reasoning 55.81%	Not reported	Unknown
Claude 3 Sonnet	47.9%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-03-04 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
InternVL-Chat-V1.2	47.7%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-02-22 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 50.93% geometry problem solving 61.06% math word problem 30.65% textbook question answering 48.1% visual question answering 44.69% algebraic reasoning 52.31% arithmetic reasoning 36.54% geometry reasoning 58.16% logical reasoning 18.92% numeric commonsense 30.56% scientific reasoning 54.92% statistical reasoning 51.83%	Not reported	Unknown
Math-LLaVA-13B	46.6%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-06-25 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 37.17% geometry problem solving 57.69% math word problem 56.45% textbook question answering 51.27% visual question answering 33.52% algebraic reasoning 53.02% arithmetic reasoning 40.23% geometry reasoning 56.49% logical reasoning 16.22% numeric commonsense 33.33% scientific reasoning 49.18% statistical reasoning 43.85%	Not reported	Unknown
LLaVA-NeXT-34B	46.5%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-01-30 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Claude 3 HaikuNamed profile referenceNamed profile reference; the source configuration remains the evaluation identity	46.4%	Chain-of-thought prompting; exact API snapshot and shot count undisclosed; Source date 2024-03-04 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Closed
Gemini 1.0 Pro	45.2%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-12-06 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 47.58% geometry problem solving 40.38% math word problem 39.25% textbook question answering 61.39% visual question answering 39.11% algebraic reasoning 45.2% arithmetic reasoning 38.81% geometry reasoning 41% logical reasoning 10.81% numeric commonsense 32.64% scientific reasoning 54.92% statistical reasoning 56.81%	Not reported	Unknown
Phi-3-Vision-128K-In	44.5%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-05-21 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Phi-3.5-Vision 4.2BSource metadata links a Phi-3 report; exact Phi-3.5 vision configuration is unverified	43.9%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-04-22 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Mini-Gemini-HD (Hermes-2-Yi-34B)	43.3%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-03-27 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Qwen-VL-Plus	43.3%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-12-21 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 54.65% geometry problem solving 38.46% math word problem 31.18% textbook question answering 55.06% visual question answering 34.08% algebraic reasoning 39.15% arithmetic reasoning 32.01% geometry reasoning 39.33% logical reasoning 18.92% numeric commonsense 26.39% scientific reasoning 59.02% statistical reasoning 56.15%	Not reported	Unknown
SPHINX-MoE	42.3%	Published method: MoE 🤖; Additional checkpoint and evaluation settings unverified; Source date 2024-01-12 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 49.81% geometry problem solving 31.25% math word problem 42.47% textbook question answering 46.84% visual question answering 39.66% algebraic reasoning 31.67% arithmetic reasoning 41.64% geometry reasoning 30.54% logical reasoning 16.22% numeric commonsense 27.08% scientific reasoning 50.82% statistical reasoning 50.83%	Not reported	Unknown
Mini-Gemini (Mixtral-8x7B)	41.8%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-03-27 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
MM1-7B-MoE-Chat	40.9%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-03-14 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
MiniCPM-V-2 (2.8B)	40.6%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-04-14 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 53.16% geometry problem solving 25.96% math word problem 37.1% textbook question answering 44.3% visual question answering 39.11% algebraic reasoning 28.47% arithmetic reasoning 33.14% geometry reasoning 28.03% logical reasoning 10.81% numeric commonsense 39.58% scientific reasoning 48.36% statistical reasoning 51.83%	Not reported	Unknown
MM1-30B-Chat	39.4%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-03-14 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
SPHINX-Plus	36.8%	Published method: MoE 🤖; Additional checkpoint and evaluation settings unverified; Source date 2024-01-12 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
SPHINX (V2)	36.7%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-11-17 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 54.65% geometry problem solving 16.35% math word problem 23.12% textbook question answering 41.77% visual question answering 43.02% algebraic reasoning 20.64% arithmetic reasoning 33.43% geometry reasoning 17.57% logical reasoning 24.32% numeric commonsense 21.53% scientific reasoning 43.44% statistical reasoning 51.5%	Not reported	Unknown
MM1-7B-Chat	35.9%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-03-14 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
SPHINX-Intern2	35.5%	Published method: MoE 🤖; Additional checkpoint and evaluation settings unverified; Source date 2024-01-12 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
OmniLMM-12B	34.9%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-02-01 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 44.98% geometry problem solving 17.79% math word problem 26.88% textbook question answering 44.94% visual question answering 39.11% algebraic reasoning 23.13% arithmetic reasoning 32.29% geometry reasoning 20.92% logical reasoning 18.92% numeric commonsense 27.78% scientific reasoning 45.9% statistical reasoning 44.19%	Not reported	Unknown
Multimodal Bard	34.8%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 26.02% geometry problem solving 47.12% math word problem 29.57% textbook question answering 48.73% visual question answering 26.82% algebraic reasoning 46.49% arithmetic reasoning 28.61% geometry reasoning 47.75% logical reasoning 13.51% numeric commonsense 14.89% scientific reasoning 47.46% statistical reasoning 33%	Not reported	Unknown
LLaVA-NeXT-Vicuna-7B	34.6%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-01-30 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
PoT GPT-4 (Caption+OCR)	33.9%	Caption and OCR inputs; published prompting setup; remaining tools and sampling settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 30.11% geometry problem solving 39.42% math word problem 30.65% textbook question answering 39.87% visual question answering 31.28% algebraic reasoning 37.37% arithmetic reasoning 31.73% geometry reasoning 41% logical reasoning 18.92% numeric commonsense 20.14% scientific reasoning 44.26% statistical reasoning 37.87%	Not reported	Unknown
CoT ChatGPT (Caption+OCR)	33.2%	Caption and OCR inputs; published prompting setup; remaining tools and sampling settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 26.02% geometry problem solving 31.73% math word problem 35.48% textbook question answering 48.1% visual question answering 30.17% algebraic reasoning 32.38% arithmetic reasoning 32.29% geometry reasoning 33.05% logical reasoning 16.22% numeric commonsense 17.36% scientific reasoning 54.92% statistical reasoning 36.21%	Not reported	Unknown
CoT Claude (Caption+OCR)	33.2%	Caption and OCR inputs; published prompting setup; remaining tools and sampling settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 27.51% geometry problem solving 29.33% math word problem 36.02% textbook question answering 49.37% visual question answering 29.05% algebraic reasoning 30.96% arithmetic reasoning 32.86% geometry reasoning 30.96% logical reasoning 16.22% numeric commonsense 17.36% scientific reasoning 50.82% statistical reasoning 37.21%	Not reported	Unknown
CoT GPT4 (Caption+OCR)	33.2%	Caption and OCR inputs; published prompting setup; remaining tools and sampling settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 27.88% geometry problem solving 31.73% math word problem 31.18% textbook question answering 51.9% visual question answering 28.49% algebraic reasoning 33.45% arithmetic reasoning 30.88% geometry reasoning 32.22% logical reasoning 13.51% numeric commonsense 12.5% scientific reasoning 58.2% statistical reasoning 37.87%	Not reported	Unknown
MM1-3B-MoE-Chat	32.6%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-03-14 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
MM1-3B-Chat	32%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-03-14 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
Gemini 1.0 Nano 2	30.6%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-12-06 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 28.62% geometry problem solving 23.56% math word problem 30.65% textbook question answering 41.77% visual question answering 31.84% algebraic reasoning 27.05% arithmetic reasoning 29.75% geometry reasoning 26.78% logical reasoning 10.81% numeric commonsense 20.83% scientific reasoning 40.16% statistical reasoning 33.55%	Not reported	Unknown
LLaVA-1.5-13B	27.6%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2024-01-30 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
SPHINX (V1)	27.5%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-11-09 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 23.42% geometry problem solving 23.08% math word problem 21.51% textbook question answering 39.87% visual question answering 34.08% algebraic reasoning 25.62% arithmetic reasoning 28.05% geometry reasoning 23.43% logical reasoning 16.22% numeric commonsense 17.36% scientific reasoning 40.16% statistical reasoning 23.59%	Not reported	Unknown
Gemini 1.0 Nano 1	27.3%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-12-06 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 30.86% geometry problem solving 21.63% math word problem 23.66% textbook question answering 29.11% visual question answering 30.73% algebraic reasoning 23.84% arithmetic reasoning 25.5% geometry reasoning 21.34% logical reasoning 13.51% numeric commonsense 20.83% scientific reasoning 27.87% statistical reasoning 30.9%	Not reported	Unknown
PoT ChatGPT (Caption+OCR)	26.8%	Caption and OCR inputs; published prompting setup; remaining tools and sampling settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 24.54% geometry problem solving 26.44% math word problem 23.66% textbook question answering 33.54% visual question answering 27.93% algebraic reasoning 27.76% arithmetic reasoning 26.06% geometry reasoning 28.03% logical reasoning 18.92% numeric commonsense 13.19% scientific reasoning 33.61% statistical reasoning 29.9%	Not reported	Unknown
SPHINX-Tiny	26.4%	Published method: MoE 🤖; Additional checkpoint and evaluation settings unverified; Source date 2024-01-12 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering Not reported geometry problem solving Not reported math word problem Not reported textbook question answering Not reported visual question answering Not reported algebraic reasoning Not reported arithmetic reasoning Not reported geometry reasoning Not reported logical reasoning Not reported numeric commonsense Not reported scientific reasoning Not reported statistical reasoning Not reported	Not reported	Unknown
LLaVA (LLaMA-2-13B)	26.1%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 26.77% geometry problem solving 29.33% math word problem 16.13% textbook question answering 32.28% visual question answering 26.26% algebraic reasoning 27.31% arithmetic reasoning 20.06% geometry reasoning 28.83% logical reasoning 24.32% numeric commonsense 18.31% scientific reasoning 37.29% statistical reasoning 25.08%	Not reported	Unknown
InstructBLIP (Vicuna-7B)	25.3%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 23.05% geometry problem solving 20.67% math word problem 18.28% textbook question answering 32.28% visual question answering 35.2% algebraic reasoning 21.77% arithmetic reasoning 27.14% geometry reasoning 20.72% logical reasoning 18.92% numeric commonsense 20.42% scientific reasoning 33.05% statistical reasoning 23.1%	Not reported	Unknown
LLaVAR	25.2%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 21.93% geometry problem solving 25% math word problem 16.67% textbook question answering 34.81% visual question answering 30.73% algebraic reasoning 24.2% arithmetic reasoning 22.1% geometry reasoning 23.01% logical reasoning 13.51% numeric commonsense 15.28% scientific reasoning 42.62% statistical reasoning 21.93%	Not reported	Unknown
LLaMA-Adapter-V2 (7B)	23.9%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 21.19% geometry problem solving 25.48% math word problem 11.29% textbook question answering 32.28% visual question answering 31.84% algebraic reasoning 26.33% arithmetic reasoning 20.4% geometry reasoning 24.27% logical reasoning 24.32% numeric commonsense 13.89% scientific reasoning 29.51% statistical reasoning 18.27%	Not reported	Unknown
miniGPT4 (LLaMA-2-7B)	23.1%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 18.59% geometry problem solving 25.96% math word problem 13.44% textbook question answering 30.38% visual question answering 30.17% algebraic reasoning 28.11% arithmetic reasoning 20.96% geometry reasoning 24.69% logical reasoning 16.22% numeric commonsense 16.67% scientific reasoning 25.41% statistical reasoning 17.94%	Not reported	Unknown
mPLUG-Owl (LLaMA-7B)	22.2%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 22.68% geometry problem solving 23.56% math word problem 10.22% textbook question answering 27.22% visual question answering 27.93% algebraic reasoning 23.62% arithmetic reasoning 19.17% geometry reasoning 23.87% logical reasoning 13.51% numeric commonsense 12.68% scientific reasoning 26.27% statistical reasoning 21.45%	Not reported	Unknown
IDEFICS (9B-Instruct)	19.8%	Published method: LMM 🖼️; Additional checkpoint and evaluation settings unverified; Source date 2023-10-03 (compiler field) Pinned aggregate Source metadata Original result Published task and skill scores figure question answering 21.56% geometry problem solving 21.15% math word problem 6.45% textbook question answering 25.95% visual question answering 24.02% algebraic reasoning 22.14% arithmetic reasoning 15.04% geometry reasoning 19.82% logical reasoning 18.92% numeric commonsense 9.86% scientific reasoning 24.58% statistical reasoning 18.15%	Not reported	Unknown

Reference baselines, outside the model table: Human Performance*: 60.3%; Random Chance: 17.86%. Missing values remain unreported. These results use different evaluation setups and do not enter overall or category rankings.

How to read this leaderboard

Scores are published accuracy percentages on a 0-100 scale. Missing task and skill scores stay empty; the importer does not calculate a replacement headline score. Compare each result with its prompting, tools, images, and compute settings.

Operator receipt: 78 sourced rows are currently displayable on this page; the highest published score among these 78 model or configuration records is DreamPRM (o4-mini) at 85.20%, which does not establish a market leader.

Honest limit: The website compiles results from multiple publishers and evaluation setups. Exact checkpoints, sample counts, and sampling settings are often undisclosed. Three named profile links are navigation references; they do not establish an exact canonical model evaluation. The testmini and test tables are separate.

MathVista project and leaderboard Pinned MathVista result compilation MathVista paper Website compilation license declaration CC BY-SA 4.0

How to read the MathVista testmini results

This table contains 78 model or configuration records from the MathVista website compilation for testmini. The split has 1,000 examples; the aggregate files do not state a separate completed-example count for each record. The pinned source commit is dated June 9, 2025, and the source files were observed on October 15, 2026.

Each score belongs to the published configuration shown below it. Prompting, tools, image processing, and test-time compute can change the result. We keep the two splits separate and exclude these scores from overall and category rankings.

Reference baselines are outside the model table: Human Performance*: 60.3%; Random Chance: 17.86%.

MathVista website compilation, CC BY-SA 4.0. Original result sources remain credited in each record. This adaptation retains CC BY-SA 4.0. Changes: Selected aggregate scores and adjacent metadata; converted numeric strings to numbers and missing-score markers to null; separated testmini from test; added configuration notes and three named profile references.

Snapshot

80 source records78 model or configuration records1,000 examples in testmini3 named profile referencesDisplay only

MathVista project and leaderboard Pinned MathVista result compilation MathVista paper Website compilation license declaration CC BY-SA 4.0

The compilation contains different evaluation setups; the highest reported value does not establish a comparable model ranking. The table retains each source configuration and its reported value.

78 model or configuration records are shown for MathVista. The benchmark falls in the Multimodal & Grounded category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. MathVista is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About MathVista

Year

2023

Tasks

1,000 examples in the testmini split

Format

Published accuracy (%)

Difficulty

Mathematical reasoning in visual contexts

The testmini split contains 1,000 examples. The pinned website compilation was committed on June 9, 2025 and observed on October 1, 2026. Source record counts describe published configurations, not distinct models or confirmed complete evaluation runs.

MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts Public benchmark source

Freshness and provenance

Version

testmini; source commit a9daca33c6b3

Refresh cadence

Pinned website compilation

Staleness state

Historical source configurations

Question availability

Public upstream; no task contents mirrored

Historical source configurationsDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

Questions

What does MathVista measure?

Which source row has the highest reported MathVista value?

DreamPRM (o4-mini) has the highest reported MathVista value at 85.20%. The compilation contains different evaluation setups; the highest reported value does not establish a comparable model ranking. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on MathVista?

The June 9, 2025 snapshot contains 78 published model or configuration records.

Last updated: June 9, 2025 · mirrored from the public benchmark leaderboard

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 5,600+ readers.

One email each week. Unsubscribe anytime.