All General Coding Reasoning Multimodal Safety Specialized Agentic Language Long Context Speed Text-to-Image Image Editing Text-to-Video Image-to-Video
BFCL83 BrowseComp OSWorld-Verified TAU-bench18 TAU-bench Airline22 TAU-bench Banking Knowledge27 TAU-bench Retail23 TAU-bench Telecom18 Tau Banking (Artificial Analysis)209 Tau2 Telecom (Artificial Analysis)439 Terminal-Bench 2.1241 Terminal-Bench Hard431 ToolBench WebArena

Agentic

Composite performance across 10 benchmarks · Normalized 0–100 · At least 5 scores required · Ranked by average performance

# Model Provider BFCL TAU-bench TAU-bench A... TAU-bench B... TAU-bench R... TAU-bench T... Tau Banking... Tau2 Teleco... Terminal-Be... Terminal-Be... Avg Coverage Price Value
1 Anthropic: Claude Opus 4.5 Anthropic 79.1% 85.3% 84.0% 24.7% 79.6% 92.3% — — — —
81.6
6/10 $30.00/M 2.7
2 OpenAI: GPT-5.2 OpenAI — 84.8% 83.0% 32.2% 81.6% 89.7% — — — —
81.1
5/10 $15.75/M 5.2
3 Anthropic: Claude Sonnet 4.5 Anthropic 73.0% 84.7% 72.0% 25.3% 86.2% 98.0% — — — —
80.5
6/10 $18.00/M 4.5
4 Qwen: Qwen3.5 397B A17B Alibaba — 87.9% 81.5% 9.8% 84.4% 97.8% — — — —
77.0
5/10 $4.10/M 18.8
5 Gemini 3 Flash Google — 83.5% 82.5% 27.3% 76.8% 91.2% — — — —
74.5
5/10 — —
6 Qwen: Qwen3 Max ThinkingHistorical Alibaba — 82.2% 69.0% — 79.4% 98.2% — 83.6% — 24.2%
72.5
6/10 $4.68/M 15.5
7 gemini-3-pro Google — 82.5% 80.5% 18.0% 75.9% 91.0% — — — —
68.0
5/10 — —
8 Google: Gemini 3.1 Pro Preview Google — — — 26.0% — — 21.4% 95.6% 73.8% 53.8%
67.2
5/10 $14.00/M 4.8
9 Z.ai: GLM 5 Z.ai — 81.0% 82.5% 9.8% 73.7% 86.8% — — — —
61.5
5/10 $2.52/M 24.4
10 Qwen: Qwen3 MaxHistorical Alibaba — 72.0% 59.5% — 72.2% 84.2% — 74.3% — 20.5%
52.1
6/10 $7.20/M 7.2
11 Cohere: Command A Cohere 31.5% — — — — — 6.0% 80.7% 22.8% 25.0%
39.1
5/10 $12.50/M 3.1
12 MoonshotAI: Kimi K2 0711 Moonshot AI — 64.3% 56.5% — 70.6% 65.8% — 61.1% — 15.9%
37.8
6/10 $2.87/M 13.2
13 Google: Gemini 2.5 ProRetiring Google — — — 13.7% — — 9.7% 54.1% 28.5% 26.5%
30.6
5/10 $11.25/M 2.7
14 OpenAI: GPT-4.1 OpenAI — 54.7% 56.0% — 74.0% 34.0% — 47.1% — 13.6%
24.1
6/10 $10.00/M 2.4
15 OpenAI: GPT-4.1 Mini OpenAI — 53.0% 48.7% — 61.4% 48.9% 5.4% 52.9% 10.1% 7.6%
13.7
8/10 $2.00/M 6.8

Anthropic: Claude Opus 4.5 leads on raw performance, but Z.ai: GLM 5 delivers 804% more performance per dollar. That extra 20.1 points costs $1.37/M per point.

Access this data

JSON GET /api/v1 
[
  {
    "rank": 1,
    "model": "Anthropic: Claude Opus 4.5",
    "provider": "Anthropic",
    "avg_score": 81.6,
    "cost": 30.0,
    "cost_unit": "million_tokens",
    "value": 2.7
  },
  {
    "rank": 2,
    "model": "OpenAI: GPT-5.2",
    "provider": "OpenAI",
    "avg_score": 81.1,
    "cost": 15.75,
    "cost_unit": "million_tokens",
    "value": 5.2
  },
  {
    "rank": 3,
    "model": "Anthropic: Claude Sonnet 4.5",
    "provider": "Anthropic",
    "avg_score": 80.5,
    "cost": 18.0,
    "cost_unit": "million_tokens",
    "value": 4.5
  }
]
Curl 
$ curl -s https://rightrank.ai/api/v1/rankings/agentic.json | jq '.[:3]'
CLI brew install rightrank 
$ rightrank top agentic --limit 5
$ rightrank compare claude-opus-4.6 gpt-5.4-pro --category agentic
$ rightrank value agentic  # best performance per dollar
Claude Code Skill Skill 
# In Claude Code, use the RightRank skill:
/rightrank best model for agentic
/rightrank compare top 5 by value
/rightrank cheapest model scoring above 80 on agentic