All General Coding Reasoning Multimodal Safety Specialized Agentic Language Long Context Speed Text-to-Image Image Editing Text-to-Video Image-to-Video
Chatbot Arena Vision87 MMMU MathVista VQAv2

Multimodal

Composite performance across 1 benchmarks · Normalized 0–100 · At least 1 scores required · Ranked by average performance · Top 50 of 87 eligible models

# Model Provider Arena Vision Avg Coverage Price Value
1 gemini-3-pro Google 1,330
100.0
1/1 — —
2 Google: Gemini 2.5 ProRetiring Google 1,263
87.0
1/1 $11.25/M 7.7
3 gemini-2.5-flash-preview-09-2025 Google 1,248
84.3
1/1 — —
5 Qwen: Qwen3 VL 235B A22B Instruct Alibaba 1,241
82.9
1/1 $2.11/M 39.3
6 chatgpt-4o-latest-20250326 OpenAI 1,241
82.8
1/1 — —
7 Google: Gemini 2.5 FlashRetiring Google 1,234
81.5
1/1 $2.80/M 29.1
8 OpenAI: GPT-5.1 OpenAI 1,233
81.4
1/1 $11.25/M 7.2
4 ↳GPT-5.1 (high) OpenAI 1,248
84.1
1/1 $11.25/M 7.5
9 gpt-5-chat OpenAI 1,230
80.7
1/1 — —
10 ernie-5.0-preview-1120 Baidu 1,224
79.5
1/1 — —
11 hunyuan-vision-1.5-thinking Tencent 1,217
78.3
1/1 — —
12 o3-2025-04-16 OpenAI 1,212
77.3
1/1 — —
13 gpt-4.1-2025-04-14 OpenAI 1,212
77.2
1/1 — —
14 qwen-vl-max-2025-08-13 Alibaba 1,212
77.2
1/1 — —
15 GPT-5 (high) OpenAI 1,208
76.6
1/1 $11.25/M 6.8
16 Qwen: Qwen3 VL 235B A22B ThinkingHistorical Alibaba 1,208
76.5
1/1 $4.40/M 17.4
17 Grok 4 SpaceXAI 1,206
76.2
1/1 $18.00/M 4.2
18 gemini-2.5-flash-lite-preview-09-2025-no-thinking Google 1,205
75.8
1/1 — —
19 GPT-5 mini (high) OpenAI 1,201
75.1
1/1 $2.25/M 33.4
21 gpt-4.5-preview-2025-02-27 OpenAI 1,194
73.7
1/1 — —
22 o4-mini-2025-04-16 OpenAI 1,193
73.6
1/1 — —
24 gemini-2.5-flash-lite-preview-06-17-thinking Google 1,188
72.7
1/1 — —
25 step-1o-turbo-202506 Stepfun 1,185
72.0
1/1 — —
26 hunyuan-large-vision Tencent 1,182
71.6
1/1 — —
27 gpt-4.1-mini-2025-04-14 OpenAI 1,181
71.4
1/1 — —
28 claude-opus-4-20250514 Anthropic 1,176
70.3
1/1 — —
20 ↳claude-opus-4-20250514-thinking-16k Anthropic 1,194
73.8
1/1 — —
29 step-3 Stepfun 1,175
70.1
1/1 — —
30 Anthropic: Claude Sonnet 4Historical Anthropic 1,174
69.9
1/1 $18.00/M 3.9
23 ↳claude-sonnet-4-20250514-thinking-32k Anthropic 1,191
73.2
1/1 $18.00/M 4.1
32 mistral-medium-2508 Mistral AI 1,169
68.9
1/1 — —
33 o1-2024-12-17 OpenAI 1,167
68.6
1/1 — —
34 Google: Gemma 3 27B Google 1,164
68.0
1/1 $0.53/M 128.3
35 gemini-1.5-pro-002 Google 1,160
67.3
1/1 — —
36 GPT-5 nano (high) OpenAI 1,159
67.0
1/1 $0.45/M 148.9
37 mistral-medium-2505 Mistral AI 1,158
66.8
1/1 — —
38 Z.ai: GLM 4.5V Z.ai 1,157
66.7
1/1 $2.40/M 27.8
39 gemini-2.0-flash-001 Google 1,157
66.7
1/1 — —
40 qwen2.5-vl-32b-instruct Alibaba 1,151
65.5
1/1 — —
41 claude-3-7-sonnet-20250219 Anthropic 1,150
65.3
1/1 — —
31 ↳claude-3-7-sonnet-20250219-thinking-32k Anthropic 1,169
69.0
1/1 — —
42 llama-4-maverick-17b-128e-instruct Meta 1,143
63.9
1/1 — —
43 mistral-small-2506 Mistral AI 1,142
63.7
1/1 — —
44 gemini-1.5-flash-002 Google 1,139
63.2
1/1 — —
45 OpenAI: GPT-4o (2024-05-13) OpenAI 1,136
62.5
1/1 $20.00/M 3.1
46 mistral-small-3.1-24b-instruct-2503 Mistral AI 1,135
62.3
1/1 — —
47 claude-3-5-sonnet-20241022 Anthropic 1,124
60.4
1/1 — —
48 step-1o-vision-32k-highres Stepfun 1,121
59.7
1/1 — —
49 claude-3-5-sonnet-20240620 Anthropic 1,119
59.4
1/1 — —
50 llama-4-scout-17b-16e-instruct Meta 1,117
59.0
1/1 — —

gemini-3-pro leads on raw performance, but GPT-5 nano (high) delivers 1488900% more performance per dollar.

Access this data

JSON GET /api/v1 
[
  {
    "rank": 1,
    "model": "gemini-3-pro",
    "provider": "Google",
    "avg_score": 100.0,
    "cost": null,
    "cost_unit": "million_tokens",
    "value": null
  },
  {
    "rank": 2,
    "model": "Google: Gemini 2.5 Pro",
    "provider": "Google",
    "avg_score": 87.0,
    "cost": 11.25,
    "cost_unit": "million_tokens",
    "value": 7.7
  },
  {
    "rank": 3,
    "model": "gemini-2.5-flash-preview-09-2025",
    "provider": "Google",
    "avg_score": 84.3,
    "cost": null,
    "cost_unit": "million_tokens",
    "value": null
  }
]
Curl 
$ curl -s https://rightrank.ai/api/v1/rankings/multimodal.json | jq '.[:3]'
CLI brew install rightrank 
$ rightrank top multimodal --limit 5
$ rightrank compare claude-opus-4.6 gpt-5.4-pro --category multimodal
$ rightrank value multimodal  # best performance per dollar
Claude Code Skill Skill 
# In Claude Code, use the RightRank skill:
/rightrank best model for multimodal
/rightrank compare top 5 by value
/rightrank cheapest model scoring above 80 on multimodal