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 Text-to-Image34 Text-to-Image Arena ELO (Artificial Analysis)102

Text-to-Image

Composite performance across 2 benchmarks · Normalized 0–100 · At least 2 scores required · Ranked by average performance

# Model Provider Arena T2I AA T2I Avg Coverage Price Value
1 Google: Nano Banana Pro (Gemini 3 Pro Image) Google 1,233 1,115
91.8
2/2 $0.13/image 682.7
2 FLUX.2 [flex] Black Forest Labs 1,156 1,027
72.9
2/2 $0.050/image 1459.0
3 Seedream 4.5 ByteDance Seed 1,146 1,028
71.5
2/2 $0.040/image 1786.3
4 FLUX.2 [pro] Black Forest Labs 1,150 1,003
70.3
2/2 $0.030/image 2341.7
5 Google: Nano Banana (Gemini 2.5 Flash Image)Retiring Google 1,156 984
69.6
2/2 $0.039/image 1798.4
6 FLUX.1 Kontext [max] Black Forest Labs 1,078 929
53.2
2/2 $0.080/image 665.0
7 FLUX.1 Kontext [pro] Black Forest Labs 1,064 920
50.3
2/2 $0.040/image 1257.5
8 Qwen Image Alibaba 1,060 898
48.1
2/2 $0.035/image 1372.9
9 FLUX1.1 [pro] Black Forest Labs 1,018 872
39.5
2/2 $0.040/image 986.3
10 bagel ByteDance Seed 911 578
0.0
2/2 — —

Google: Nano Banana Pro (Gemini 3 Pro Image) leads on raw performance, but FLUX.2 [pro] delivers 243% more performance per dollar. That extra 21.5 points costs $0.005/image per point.

Access this data

JSON GET /api/v1 
[
  {
    "rank": 1,
    "model": "Google: Nano Banana Pro (Gemini 3 Pro Image)",
    "provider": "Google",
    "avg_score": 91.8,
    "cost": 0.1344,
    "cost_unit": "image",
    "value": 682.7
  },
  {
    "rank": 2,
    "model": "FLUX.2 [flex]",
    "provider": "Black Forest Labs",
    "avg_score": 72.9,
    "cost": 0.05,
    "cost_unit": "image",
    "value": 1459.0
  },
  {
    "rank": 3,
    "model": "Seedream 4.5",
    "provider": "ByteDance Seed",
    "avg_score": 71.5,
    "cost": 0.04,
    "cost_unit": "image",
    "value": 1786.3
  }
]
Curl 
$ curl -s https://rightrank.ai/api/v1/rankings/text-to-image.json | jq '.[:3]'
CLI brew install rightrank 
$ rightrank top text-to-image --limit 5
$ rightrank compare claude-opus-4.6 gpt-5.4-pro --category text-to-image
$ rightrank value text-to-image  # best performance per dollar
Claude Code Skill Skill 
# In Claude Code, use the RightRank skill:
/rightrank best model for text-to-image
/rightrank compare top 5 by value
/rightrank cheapest model scoring above 80 on text-to-image