Leaderboard
vision rankings
Updated weeklyEach leaderboard uses transparent weighted scoring, current model context, and supporting analysis to help teams interpret the results with confidence. Only full-profile entries appear in rankings; broader catalog records remain available elsewhere on the site when only source-backed metadata is currently available.
Models with complete enough metadata and scoring coverage to be meaningfully ranked in this category.
Scores combine benchmark evidence, product metadata, and cost/context signals when those fields are published.
Tracked models without full scoring remain in the directory and provider pages, but are not relied on for analytical ranking claims.
| Rank | Model | Provider | Score | Context |
|---|---|---|---|---|
| #1 | Gemini 3.1 Pro | Google DeepMind | 93 | 1,048,576 |
| #2 | GPT-4o | OpenAI | 93 | 128,000 |
| #3 | GPT-5.4 | OpenAI | 93 | 1,000,000 |
| #4 | Gemini 2.5 Pro | Google DeepMind | 92 | 1,048,576 |
| #5 | GPT-5.4 Pro | OpenAI | 92 | 1,000,000 |
| #6 | Gemini 2.5 Pro TTS | Google DeepMind | 91 | 1,048,576 |
| #7 | Gemini 3.0 Pro | Google DeepMind | 91 | 1,048,576 |
| #8 | Gemini 3.1 Flash | Google DeepMind | 91 | 1,048,576 |
| #9 | GPT-5.2 | OpenAI | 91 | 1,000,000 |
| #10 | Claude Opus 4.6 | Anthropic | 90 | 1,000,000 |
| #11 | Claude Sonnet 4.5 | Anthropic | 90 | 1,000,000 |
| #12 | Claude Sonnet 4.6 | Anthropic | 90 | 1,000,000 |
| #13 | GPT-5.2 Pro | OpenAI | 90 | 1,000,000 |
| #14 | Gemini 2.5 Flash | Google DeepMind | 89 | 1,048,576 |
| #15 | Gemini 3.0 Flash | Google DeepMind | 89 | 1,048,576 |
| #16 | Claude Sonnet 4 | Anthropic | 88 | 1,000,000 |
| #17 | GPT-5 | OpenAI | 88 | 1,000,000 |
| #18 | Claude Opus 4.1 | Anthropic | 87 | 200,000 |
| #19 | Gemini 2.0 Flash | Google DeepMind | 87 | 1,048,576 |
| #20 | Claude 3.7 Sonnet | Anthropic | 86 | 200,000 |
| #21 | Claude Opus 4 | Anthropic | 86 | 200,000 |
| #22 | Gemini 2.5 Flash Live | Google DeepMind | 85 | 1,048,576 |
| #23 | Gemini 2.5 Flash Native Audio Preview | Google DeepMind | 85 | 1,048,576 |
| #24 | Gemini 2.5 Flash-Lite | Google DeepMind | 85 | 1,048,576 |
| #25 | Gemini 3.1 Flash-Lite | Google DeepMind | 84 | 1,048,576 |
| #26 | GPT-4.1 | OpenAI | 84 | 1,048,576 |
| #27 | Mistral Large 25 | Mistral AI | 84 | 128,000 |
| #28 | GPT Image 1 | OpenAI | 83 | 32,768 |
| #29 | GPT-5 mini | OpenAI | 83 | 1,000,000 |
| #30 | Llama 4 Maverick | Meta | 83 | 1,048,576 |
| #31 | Gemini 1.5 Pro | Google DeepMind | 82 | 2,097,152 |
| #32 | GPT-5.3 Instant | OpenAI | 82 | 1,000,000 |
| #33 | Claude Haiku 4.5 | Anthropic | 80 | 200,000 |
| #34 | Gemini 2.0 Flash-Lite | Google DeepMind | 80 | 1,048,576 |
| #35 | chatgpt-image-latest | OpenAI | 79 | 32,768 |
| #36 | GPT-5.3-Codex | OpenAI | 79 | 1,000,000 |
| #37 | gpt-image-1-mini | OpenAI | 79 | 32,768 |
| #38 | Nova Pro | Amazon Web Services | 78 | 300,000 |
| #39 | Claude Opus 3 | Anthropic | 77 | 200,000 |
| #40 | Gemini 1.5 Flash | Google DeepMind | 77 | 1,048,576 |
| #41 | GPT-4.1 mini | OpenAI | 76 | 1,048,576 |
| #42 | GPT-5 nano | OpenAI | 75 | 1,000,000 |
| #43 | o3 | OpenAI | 75 | 200,000 |
| #44 | Claude Haiku 3.5 | Anthropic | 74 | 200,000 |
| #45 | Claude Sonnet 3 | Anthropic | 74 | 200,000 |
| #46 | FLUX 1.1 Pro | Black Forest Labs | 74 | 512 |
| #47 | FLUX 1.1 Pro Ultra | Black Forest Labs | 74 | 512 |
| #48 | Llama 3.2 90B Vision Instruct | Meta | 74 | 128,000 |
| #49 | o3-deep-research | OpenAI | 74 | 200,000 |
| #50 | o4-mini-deep-research | OpenAI | 74 | 200,000 |
| #51 | GPT-4o-mini | OpenAI | 73 | 128,000 |
| #52 | Command R+ 2026 | Cohere | 72 | 128,000 |
| #53 | o4-mini | OpenAI | 72 | 200,000 |
| #54 | Step3-VL-10B | StepFun | 72 | 131,072 |
| #55 | FLUX 1 Pro | Black Forest Labs | 71 | 512 |
| #56 | Llama 4 Scout | Meta | 71 | 10,485,760 |
| #57 | MiMo-VL-7B | Xiaomi | 71 | 131,072 |
| #58 | Gemini 1.5 Flash-8B | Google DeepMind | 70 | 1,048,576 |
| #59 | Nova Lite | Amazon Web Services | 70 | 300,000 |
| #60 | Llama 3.2 11B Vision Instruct | Meta | 68 | 128,000 |
| #61 | Claude Haiku 3 | Anthropic | 67 | 200,000 |
| #62 | Claude Haiku 3.5 | Anthropic | 66 | 200,000 |
| #63 | Claude Haiku 4.5 | Anthropic | 66 | 200,000 |
| #64 | Codestral | Mistral AI | 66 | 256,000 |
| #65 | Codestral Embed | Mistral AI | 66 | 32,768 |
| #66 | CogVideoX | Z.AI | 66 | 128,000 |
| #67 | CogView 4 | Z.AI | 66 | 128,000 |
| #68 | Devstral Medium 1.0 | Mistral AI | 66 | 128,000 |
| #69 | Doubao-Seed-1.6 | ByteDance / Doubao | 66 | 128,000 |
| #70 | Doubao-Seed-1.6-Flash | ByteDance / Doubao | 66 | 128,000 |
| #71 | Doubao-Seed-2.0-Code | ByteDance / Doubao | 66 | 128,000 |
| #72 | Doubao-Seed-Code | ByteDance / Doubao | 66 | 128,000 |
| #73 | ERNIE 3.5 128K | Baidu / ERNIE | 66 | 128,000 |
| #74 | ERNIE 4.0 Turbo 8K | Baidu / ERNIE | 66 | 128,000 |
| #75 | ERNIE Functions 8K | Baidu / ERNIE | 66 | 128,000 |
| #76 | ERNIE Speed 128K | Baidu / ERNIE | 66 | 128,000 |
| #77 | GLM-4.5 | Z.AI | 66 | 128,000 |
| #78 | GLM-4.5V | Z.AI | 66 | 128,000 |
| #79 | GLM-4.6 | Z.AI | 66 | 128,000 |
| #80 | GLM-4.6V | Z.AI | 66 | 128,000 |
| #81 | GLM-4.7 | Z.AI | 66 | 128,000 |
| #82 | GLM-5 | Z.AI | 66 | 128,000 |
| #83 | GLM-Image | Z.AI | 66 | 128,000 |
| #84 | GLM-OCR | Z.AI | 66 | 128,000 |
| #85 | Hunyuan Code | Tencent / Hunyuan | 66 | 128,000 |
| #86 | Hunyuan Lite | Tencent / Hunyuan | 66 | 128,000 |
| #87 | Hunyuan Standard | Tencent / Hunyuan | 66 | 128,000 |
| #88 | Hunyuan T1 | Tencent / Hunyuan | 66 | 256,000 |
| #89 | Hunyuan T1 Vision | Tencent / Hunyuan | 66 | 128,000 |
| #90 | Hunyuan TurboS | Tencent / Hunyuan | 66 | 128,000 |
| #91 | Hunyuan TurboS LongText 128K | Tencent / Hunyuan | 66 | 128,000 |
| #92 | Kimi K2 | Moonshot AI / Kimi | 66 | 131,072 |
| #93 | Kimi K2 Thinking | Moonshot AI / Kimi | 66 | 256,000 |
| #94 | Kimi K2 Turbo Preview | Moonshot AI / Kimi | 66 | 256,000 |
| #95 | Kimi K2.5 | Moonshot AI / Kimi | 66 | 256,000 |
| #96 | Llama 3.3 70B Instruct | Meta | 66 | 128,000 |
| #97 | Magistral Medium 1.2 | Mistral AI | 66 | 128,000 |
| #98 | MiniMax-M1 | MiniMax | 66 | 204,800 |
| #99 | MiniMax-M2 | MiniMax | 66 | 204,800 |
| #100 | MiniMax-M2.1 | MiniMax | 66 | 204,800 |
| #101 | MiniMax-M2.1-highspeed | MiniMax | 66 | 204,800 |
| #102 | MiniMax-M2.5 | MiniMax | 66 | 204,800 |
| #103 | MiniMax-M2.5-highspeed | MiniMax | 66 | 204,800 |
| #104 | MiniMax-Text-01 | MiniMax | 66 | 204,800 |
| #105 | MiniMax-VL-01 | MiniMax | 66 | 204,800 |
| #106 | Mistral Embed | Mistral AI | 66 | 32,768 |
| #107 | Mistral Large 3 | Mistral AI | 66 | 128,000 |
| #108 | Mistral Medium 3.1 | Mistral AI | 66 | 128,000 |
| #109 | Mistral Moderation | Mistral AI | 66 | 32,768 |
| #110 | Mistral Small 3.1 | Mistral AI | 66 | 128,000 |
| #111 | Mistral Small 3.2 Open | Mistral AI | 66 | 128,000 |
| #112 | Pixtral 12B | Mistral AI | 66 | 131,072 |
| #113 | Pixtral Large | Mistral AI | 66 | 131,072 |
| #114 | Vidu Q1 | Z.AI | 66 | 128,000 |
| #115 | Voxtral Mini Open | Mistral AI | 66 | 131,072 |
| #116 | Voxtral Mini Transcribe | Mistral AI | 66 | 131,072 |
| #117 | Voxtral Small Open | Mistral AI | 66 | 131,072 |
| #118 | Claude Haiku 3 | Anthropic | 65 | 200,000 |
| #119 | Claude Sonnet 3 | Anthropic | 65 | 200,000 |
| #120 | Claude Sonnet 4 | Anthropic | 65 | 1,000,000 |
| #121 | Command A | Cohere | 65 | 256,000 |
| #122 | Command A Reasoning | Cohere | 65 | 128,000 |
| #123 | Command A Translate | Cohere | 65 | 128,000 |
| #124 | Command A Vision | Cohere | 65 | 128,000 |
| #125 | Command R+ | Cohere | 65 | 128,000 |
| #126 | Command R7B | Cohere | 65 | 128,000 |
| #127 | DeepSeek-Coder-V2 | DeepSeek | 65 | 128,000 |
| #128 | DeepSeek-Math-V2 | DeepSeek | 65 | 128,000 |
| #129 | DeepSeek-R1 | DeepSeek | 65 | 128,000 |
| #130 | DeepSeek-R1-Distill-Llama-70B | DeepSeek | 65 | 128,000 |
| #131 | DeepSeek-V2.5 | DeepSeek | 65 | 128,000 |
| #132 | DeepSeek-V3 | DeepSeek | 65 | 128,000 |
| #133 | DeepSeek-V3.1 | DeepSeek | 65 | 128,000 |
| #134 | DeepSeek-V3.1-Base | DeepSeek | 65 | 128,000 |
| #135 | DeepSeek-V3.2 | DeepSeek | 65 | 128,000 |
| #136 | DeepSeek-V3.2-Exp | DeepSeek | 65 | 128,000 |
| #137 | Devstral 2 Open | Mistral AI | 65 | 128,000 |
| #138 | Devstral Small 2 | Mistral AI | 65 | 128,000 |
| #139 | Embed 4 | Cohere | 65 | 128,000 |
| #140 | ERNIE 4.5 Turbo 32K | Baidu / ERNIE | 65 | 32,768 |
| #141 | Grok 3 | xAI | 65 | 131,072 |
| #142 | Grok 3 Mini | xAI | 65 | 131,072 |
| #143 | Grok 4 | xAI | 65 | 256,000 |
| #144 | Grok 4 Fast Reasoning | xAI | 65 | 131,072 |
| #145 | grok-image | xAI | 65 | 131,072 |
| #146 | Jamba 3B | AI21 Labs | 65 | 256,000 |
| #147 | Jamba Large | AI21 Labs | 65 | 256,000 |
| #148 | Jamba Large 1.6 | AI21 Labs | 65 | 256,000 |
| #149 | Jamba Mini | AI21 Labs | 65 | 256,000 |
| #150 | Jamba Mini 1.6 | AI21 Labs | 65 | 256,000 |
| #151 | Jamba Mini 1.7 | AI21 Labs | 65 | 256,000 |
| #152 | Magistral Small 1.2 Open | Mistral AI | 65 | 128,000 |
| #153 | Ministral 3 14B Open | Mistral AI | 65 | 128,000 |
| #154 | Ministral 3 3B Open | Mistral AI | 65 | 128,000 |
| #155 | Ministral 3 8B Open | Mistral AI | 65 | 128,000 |
| #156 | Mistral Large 3 Open | Mistral AI | 65 | 128,000 |
| #157 | Mistral Nemo 12B | Mistral AI | 65 | 128,000 |
| #158 | Mistral OCR 2505 | Mistral AI | 65 | 32,768 |
| #159 | Phi-3-vision-128k-instruct | Microsoft | 65 | 128,000 |
| #160 | Phi-3.5-mini-instruct | Microsoft | 65 | 131,072 |
| #161 | Phi-3.5-MoE-instruct | Microsoft | 65 | 131,072 |
| #162 | Phi-3.5-vision-instruct | Microsoft | 65 | 131,072 |
| #163 | Phi-4-mini-flash-reasoning | Microsoft | 65 | 131,072 |
| #164 | Phi-4-mini-instruct | Microsoft | 65 | 131,072 |
| #165 | Phi-4-multimodal-instruct | Microsoft | 65 | 131,072 |
| #166 | Phi-4-reasoning | Microsoft | 65 | 131,072 |
| #167 | Phi-4-reasoning-plus | Microsoft | 65 | 131,072 |
| #168 | Phi-4-reasoning-vision-15B | Microsoft | 65 | 131,072 |
| #169 | Qwen2.5-1.5B-Instruct | Alibaba Qwen | 65 | 131,072 |
| #170 | Qwen2.5-14B-Instruct | Alibaba Qwen | 65 | 131,072 |
| #171 | Qwen2.5-32B-Instruct | Alibaba Qwen | 65 | 131,072 |
| #172 | Qwen2.5-3B-Instruct | Alibaba Qwen | 65 | 131,072 |
| #173 | Qwen2.5-72B-Instruct | Alibaba Qwen | 65 | 131,072 |
| #174 | Qwen2.5-7B-Instruct | Alibaba Qwen | 65 | 131,072 |
| #175 | Qwen2.5-Max | Alibaba Qwen | 65 | 131,072 |
| #176 | Qwen2.5-VL-72B-Instruct | Alibaba Qwen | 65 | 131,072 |
| #177 | Qwen2.5-VL-7B-Instruct | Alibaba Qwen | 65 | 131,072 |
| #178 | Qwen3-Coder-Next | Alibaba Qwen | 65 | 131,072 |
| #179 | Qwen3.5-0.8B | Alibaba Qwen | 65 | 131,072 |
| #180 | Qwen3.5-122B-A10B | Alibaba Qwen | 65 | 131,072 |
| #181 | Qwen3.5-27B | Alibaba Qwen | 65 | 131,072 |
| #182 | Qwen3.5-2B | Alibaba Qwen | 65 | 131,072 |
| #183 | Qwen3.5-35B-A3B | Alibaba Qwen | 65 | 131,072 |
| #184 | Qwen3.5-397B-A17B | Alibaba Qwen | 65 | 131,072 |
| #185 | Qwen3.5-4B | Alibaba Qwen | 65 | 131,072 |
| #186 | Qwen3.5-9B | Alibaba Qwen | 65 | 131,072 |
| #187 | DeepSeek-OCR | DeepSeek | 64 | 16,384 |
| #188 | DeepSeek-OCR-2 | DeepSeek | 64 | 16,384 |
| #189 | DeepSeek-VL2-Small | DeepSeek | 64 | 16,384 |
| #190 | image-01 | MiniMax | 64 | 8,192 |
| #191 | image-01-live | MiniMax | 64 | 8,192 |
| #192 | Janus-Pro-7B | DeepSeek | 64 | 16,384 |
| #193 | MiniMax-Speech-02 | MiniMax | 64 | 8,192 |
| #194 | music-2.0 | MiniMax | 64 | 8,192 |
| #195 | NextStep-1.1 | StepFun | 64 | 512 |
| #196 | o1 | OpenAI | 64 | 200,000 |
| #197 | Phi-4 | Microsoft | 64 | 16,384 |
| #198 | Claude Opus 3 | Anthropic | 63 | 200,000 |
| #199 | Claude Opus 4 | Anthropic | 63 | 200,000 |
| #200 | phi-1 | Microsoft | 63 | 4,096 |
| #201 | phi-1_5 | Microsoft | 63 | 4,096 |
| #202 | phi-2 | Microsoft | 63 | 4,096 |
| #203 | Phi-3-medium-4k-instruct | Microsoft | 63 | 4,096 |
| #204 | Phi-3-mini-4k-instruct | Microsoft | 63 | 4,096 |
| #205 | Phi-tiny-MoE-instruct | Microsoft | 63 | 4,096 |
| #206 | Llama Guard 4 12B | Meta | 62 | 131,072 |
| #207 | Llama 3.1 405B Instruct | Meta | 61 | 128,000 |
| #208 | Llama 3.1 70B Instruct | Meta | 61 | 128,000 |
| #209 | gpt-oss-120b | OpenAI | 59 | 131,072 |
| #210 | MiMo-Audio-7B | Xiaomi | 59 | 131,072 |
| #211 | o1-mini | OpenAI | 59 | 128,000 |
| #212 | gpt-audio | OpenAI | 58 | 128,000 |
| #213 | gpt-realtime | OpenAI | 58 | 128,000 |
| #214 | Llama Guard 3 11B Vision | Meta | 57 | 131,072 |
| #215 | Sonar Pro | Perplexity | 55 | 200,000 |
| #216 | Llama 3.1 8B Instruct | Meta | 54 | 128,000 |
| #217 | Sonar Reasoning Pro | Perplexity | 54 | 200,000 |
| #218 | Step-3.5-Flash | StepFun | 54 | 131,072 |
| #219 | Nova Micro | Amazon Web Services | 53 | 128,000 |
| #220 | Step-Audio-R1.1 | StepFun | 53 | 131,072 |
| #221 | gpt-audio-mini | OpenAI | 52 | 128,000 |
| #222 | gpt-oss-20b | OpenAI | 52 | 131,072 |
| #223 | gpt-realtime-mini | OpenAI | 52 | 128,000 |
| #224 | Sonar | Perplexity | 52 | 128,000 |
| #225 | Llama 3.2 1B Instruct | Meta | 51 | 128,000 |
| #226 | Llama 3.2 3B Instruct | Meta | 51 | 128,000 |
| #227 | Sonar Deep Research | Perplexity | 51 | 200,000 |
| #228 | Code Llama 70B Instruct | Meta | 46 | 8,192 |
| #229 | Meta Llama 3 70B Instruct | Meta | 46 | 8,192 |
| #230 | GPT-4o mini Transcribe | OpenAI | 44 | 128,000 |
| #231 | GPT-4o mini TTS | OpenAI | 44 | 128,000 |
| #232 | GPT-4o Transcribe | OpenAI | 44 | 128,000 |
| #233 | Code Llama 34B Instruct | Meta | 43 | 8,192 |
| #234 | Meta Llama 3 8B Instruct | Meta | 43 | 8,192 |
| #235 | morph-v3-fast-apply | Morph | 43 | 128,000 |
| #236 | warpgrep-v2 | Morph | 42 | 128,000 |
| #237 | LFM2-24B-A2B | Liquid AI | 41 | 32,768 |
| #238 | flash-compact | Morph | 39 | 200,000 |
| #239 | LFM2-8B-A1B | Liquid AI | 36 | 32,768 |
| #240 | LFM2.5-1.2B-Instruct | Liquid AI | 32 | 131,072 |
| #241 | LFM2.5-1.2B-Thinking | Liquid AI | 32 | 131,072 |
| #242 | LFM2-2.6B | Liquid AI | 31 | 32,768 |
| #243 | pplx-embed-v1-4b | Perplexity | 31 | 8,192 |
| #244 | pplx-embed-v1-0.6b | Perplexity | 30 | 8,192 |
| #245 | Prompt Guard 86M | Meta | 29 | 512 |
Why #1: Gemini 3.1 Pro
Google's Gemini 3.1 Pro, designed for complex tasks where simple answers aren't enough. Released Feb 2026 with enhanced reasoning and multimodal capabilities.
This model clears the current full-profile threshold for leaderboard methodology.
Why #2: GPT-4o
A broadly capable multimodal model optimized for production chat, agentic workflows, and voice experiences.
This model clears the current full-profile threshold for leaderboard methodology.
Why #3: GPT-5.4
OpenAI's GPT-5.4, the most capable and efficient frontier model for professional work. First general-purpose model with native computer-use capabilities. Combines industry-leading coding from GPT-5.3-Codex with improved agentic workflows.
This model clears the current full-profile threshold for leaderboard methodology.