Atlas

Benchmarks

Visa Agentic Commerce Benchmark

The Visa Agentic Commerce Benchmark (Visa Research) tests whether an AI shopping assistant can work out what a shopper actually needs. In each of 15 episodes the assistant talks with a simulated customer who reveals requirements only when asked, and researches a fixed market with tools under limits on turns, tool calls and tokens. A run passes when the purchase meets every requirement in the answer key and each requirement came out during the session, or when the assistant rightly ends without a purchase because nothing fits. Each model runs every episode three times, and the score is its share of 45 runs passed. One episode is public and 14 are private. Visa says the two leaders' confidence intervals overlap those of the next eleven models.

Models
22
Top score
82.2
Leader
Gemini 4 Argon
Updated
Nov 5, 2026
Resolution
~4 mo
Useful
Saturation
~Aug 2027
Projected
Signal
38thpercentile
205 benchmarks from 0.8 to 3.6219 benchmarks from 3.6 to 6.4134 benchmarks from 6.4 to 9.393 benchmarks from 9.3 to 12.155 benchmarks from 12.1 to 14.925 benchmarks from 14.9 to 17.713 benchmarks from 17.7 to 20.55 benchmarks from 20.5 to 23.37 benchmarks from 23.3 to 26.27 benchmarks from 26.2 to 29.01 benchmark from 29.0 to 31.83 benchmarks from 31.8 to 34.66 benchmarks from 34.6 to 37.43 benchmarks from 37.4 to 40.21 benchmark from 40.2 to 43.11 benchmark from 45.9 to 48.71 benchmark from 65.6 to 68.4

480th strongest of 779

Signal-to-noise +4.70

Difficulty
60thpercentile
1 benchmark from 5.9 to 14.81 benchmark from 23.6 to 32.52 benchmarks from 32.5 to 41.31 benchmark from 41.3 to 50.22 benchmarks from 50.2 to 59.08 benchmarks from 59.0 to 67.97 benchmarks from 67.9 to 76.819 benchmarks from 76.8 to 85.623 benchmarks from 85.6 to 94.541 benchmarks from 94.5 to 103.331 benchmarks from 103.3 to 112.239 benchmarks from 112.2 to 121.148 benchmarks from 121.1 to 129.973 benchmarks from 129.9 to 138.859 benchmarks from 138.8 to 147.6118 benchmarks from 147.6 to 156.5127 benchmarks from 156.5 to 165.3113 benchmarks from 165.3 to 174.235 benchmarks from 174.2 to 183.117 benchmarks from 183.1 to 191.99 benchmarks from 191.9 to 200.81 benchmark from 200.8 to 209.64 benchmarks from 209.6 to 218.5

307th hardest of 779

Capability needed 156.4

22 dated models · 6 frontier modelsClosed weightsOpen weights
#ModelSettingScoreSource
1Gemini 4 Argonnot reported
82.2
2Muse Wondernot reported
82.2
3Claude Fable 5.1not reported
75.6
4GLM-5.3not reported
66.7
5Claude Sonnet 5.5not reported
64.4
6GPT-6 Astranot reported
64.4
7Muse Spark 1.3not reported
64.4
8Claude Opus 5.5not reported
62.2
9GPT-6.1 Solnot reported
62.2
10Grok 4.7not reported
60.0
11Qwen3.8 Max (0902)not reported
60.0
12DeepSeek V4.1 Flash 1017not reported
57.8
13Beam 1.0 501B A23Bnot reported
55.6
14Gemini 3.8 Flashnot reported
53.3
15GPT-6 Lunanot reported
53.3
16Kimi K3.1 Pronot reported
53.3
17GPT-6.1 Terranot reported
51.1
18MiniMax M3not reported
48.9
19DeepSeek-V4-Pronot reported
42.2
20Gemini 3.1 Pro Previewnot reported
35.6
21MiniMax M3.1 Flashnot reported
33.3
22Inklingnot reported
28.9
Loading Atlas data…