Skip to content
Fifth Order
← Benchmarks

Agentic finance · Fifth Order

Financial Workflows Benchmark

Models complete multi-step financial workflows: gathering evidence, interpreting statements and turning analysis into a usable answer. Evaluation considers whether the requested work is completed, calculations reconcile and conclusions remain supported by the available information. The score summarizes task performance on a 0–100 scale; higher values indicate stronger completion across the workflow collection.

Model results

38 configurations
RankModelScore
  1. 1
    xAIGrok 4.6 xHigh reasoning
    88.2%
  2. 2
    OpenAIGPT 5.6 Luna Max reasoning
    87.2%
  3. 3
    ZhipuGLM 5.3 Max reasoning
    82.2%
  4. 4
    DeepSeekDeepSeek V4 Flash Max reasoning
    81.2%
  5. 5
    AnthropicClaude Fable 5.1 Adaptive/Max
    77.2%
  6. 6
    OpenAIGPT 6 Luna Max reasoning
    76.2%
  7. 7
    MetaMuse Glimmer 30B xHigh reasoning
    75.2%
  8. 8
    AnthropicClaude Opus 5 Adaptive/Max
    71.2%
  9. 9
    GoogleGemini 3.8 Flash High reasoning
    70.2%
  10. 10
    OpenAIGPT 5.6 Sol Max reasoning
    65.2%
  11. 11
    TencentHy3 High reasoning
    64.2%
  12. 12
    OpenAIGPT 6 Astra Max reasoning
    60.2%
  13. 13
    MetaMuse Spark 1.2 xHigh reasoning
    59.2%
  14. 14
    MistralMistral Large 3 Not specified by publisher
    58.2%
  15. 15
    xAIGrok 4.7 xHigh reasoning
    54.2%
  16. 16
    AnthropicClaude Sonnet 5 Adaptive/Max
    53.2%
  17. 17
    DeepSeekDeepSeek V4.1 Flash 1017 Max reasoning
    48.2%
  18. 18
    ◇Beam 1.0 Max reasoning
    47.2%
  19. 19
    MetaMuse Wonder 1.0 Max reasoning
    43.2%
  20. 20
    ZhipuGLM 5.3 Flash Max reasoning
    42.2%
  21. 21
    MoonshotKimi K2.7 Code Max reasoning
    41.2%
  22. 22
    MetaMuse Spark 1.3 Max reasoning
    37.2%
  23. 23
    MistralMistral Large 4 High reasoning
    36.2%
  24. 24
    AnthropicClaude Fable 5 Adaptive/Max
    31.2%
  25. 25
    OpenAIGPT 5.6 Terra xHigh reasoning
    30.2%
  26. 26
    AnthropicClaude Sonnet 5.5 Adaptive/Max
    26.2%
  27. 27
    AlibabaQwen 3.8 Max xHigh reasoning
    25.2%
  28. 28
    ◇Inkling Max reasoning
    24.2%
  29. 29
    GoogleGemini 4 Argon High reasoning
    20.2%
  30. 30
    GoogleGemini 3.7 Flash High reasoning
    19.2%
  31. 31
    OpenAIGPT 6 Sol Max reasoning
    14.2%
  32. 32
    DeepSeekDeepSeek V4 Pro Max reasoning
    13.2%
  33. 33
    AnthropicClaude Opus 5.5 Adaptive/Max
    9.2%
  34. 34
    TencentHy4 Preview High reasoning
    8.2%
  35. 35
    GoogleGemini 3.1 Pro High reasoning
    7.2%
  36. 36
    OpenAIGPT 6.1 Sol Max reasoning
    3.2%
  37. 37
    MoonshotKimi K3 Max reasoning
    2.2%
  38. 38
    NVIDIANemotron 3 Ultra Not specified by publisher
    1.2%
Numeric table · 38 configurations
Score · highest first
RankModel / settingScore
1
xAIGrok 4.6 xHigh reasoning
88.2%
2
OpenAIGPT 5.6 Luna Max reasoning
87.2%
3
ZhipuGLM 5.3 Max reasoning
82.2%
4
DeepSeekDeepSeek V4 Flash Max reasoning
81.2%
5
AnthropicClaude Fable 5.1 Adaptive/Max
77.2%
6
OpenAIGPT 6 Luna Max reasoning
76.2%
7
MetaMuse Glimmer 30B xHigh reasoning
75.2%
8
AnthropicClaude Opus 5 Adaptive/Max
71.2%
9
GoogleGemini 3.8 Flash High reasoning
70.2%
10
OpenAIGPT 5.6 Sol Max reasoning
65.2%
11
TencentHy3 High reasoning
64.2%
12
OpenAIGPT 6 Astra Max reasoning
60.2%
13
MetaMuse Spark 1.2 xHigh reasoning
59.2%
14
MistralMistral Large 3 Not specified by publisher
58.2%
15
xAIGrok 4.7 xHigh reasoning
54.2%
16
AnthropicClaude Sonnet 5 Adaptive/Max
53.2%
17
DeepSeekDeepSeek V4.1 Flash 1017 Max reasoning
48.2%
18
◇Beam 1.0 Max reasoning
47.2%
19
MetaMuse Wonder 1.0 Max reasoning
43.2%
20
ZhipuGLM 5.3 Flash Max reasoning
42.2%
21
MoonshotKimi K2.7 Code Max reasoning
41.2%
22
MetaMuse Spark 1.3 Max reasoning
37.2%
23
MistralMistral Large 4 High reasoning
36.2%
24
AnthropicClaude Fable 5 Adaptive/Max
31.2%
25
OpenAIGPT 5.6 Terra xHigh reasoning
30.2%
26
AnthropicClaude Sonnet 5.5 Adaptive/Max
26.2%
27
AlibabaQwen 3.8 Max xHigh reasoning
25.2%
28
◇Inkling Max reasoning
24.2%
29
GoogleGemini 4 Argon High reasoning
20.2%
30
GoogleGemini 3.7 Flash High reasoning
19.2%
31
OpenAIGPT 6 Sol Max reasoning
14.2%
32
DeepSeekDeepSeek V4 Pro Max reasoning
13.2%
33
AnthropicClaude Opus 5.5 Adaptive/Max
9.2%
34
TencentHy4 Preview High reasoning
8.2%
35
GoogleGemini 3.1 Pro High reasoning
7.2%
36
OpenAIGPT 6.1 Sol Max reasoning
3.2%
37
MoonshotKimi K3 Max reasoning
2.2%
38
NVIDIANemotron 3 Ultra Not specified by publisher
1.2%
Methodology

Performance is scored on a 0–100 scale across retrieval, calculation, synthesis and judgment. Higher scores rank first; equal scores share a rank.