# Financial Tool-Use on FinToolBench Source: [Financial Tool-Use on FinToolBench]() Collection index: [llms.txt]() Data updated: 2026-03-26T16:28:12.293Z Evaluation of a model's capability to utilize specialized tools within a financial context, measured across various success and dependency metrics (Avg, ST-SC, MT-MC, etc.). A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Financial Tool-Use | | Dataset (including any stated split) | FinToolBench | | Metric | Average Success Rate | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 31 of 31 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Average Success Rate](): Higher is better; unit: PERCENT. - [ST-SC Success Rate](): Higher is better; unit: PERCENT. - [ST-MC Success Rate](): Higher is better; unit: PERCENT. - [MT-SC Success Rate](): Higher is better; unit: PERCENT. - [MT-MC Success Rate](): Higher is better; unit: PERCENT. - [Utility Dependency Score (UD)](): Higher is better; unit: PERCENT. - [Correctness Index (CI)](): Higher is better; unit: PERCENT. - [Dependency Rate (DR)](): Higher is better; unit: PERCENT. - [Information Transfer Accuracy (ITA)](): Higher is better; unit: PERCENT. - [Key Dependency Achievement (KDA)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Average Success Rate | Result date | Report paper | | --- | --- | --- | --- | | Gemini-2.5-Pro | 62.15 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Gemini-2.5-Pro | 60.98 | Not recorded | 69c4bdb20636e5bf4afc7717 | | GPT-4o | 59.67 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Gemini-2.5-Flash | 58.35 | Not recorded | 69c4bdb20636e5bf4afc7717 | | GPT-4o | 58.26 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen3-235B-Instruct | 58.06 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Claude-4.0-Sonnet | 57.96 | Not recorded | 69c4bdb20636e5bf4afc7717 | | FinTool-Qwen3-14B | 57.95 | Not recorded | 69c4bdb20636e5bf4afc7717 | | FinTool-Qwen3-14B | 57.77 | Not recorded | 69c4bdb20636e5bf4afc7717 | | FinTool-Qwen3-8B | 57.64 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen2.5-14B-Instruct | 57.23 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Claude-4.0-Sonnet | 55.93 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen2.5-14B-Instruct | 55.75 | Not recorded | 69c4bdb20636e5bf4afc7717 | | FinTool-Qwen3-8B | 55.68 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Deepseek-V3.1-Terminus | 55.32 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen2.5-72B-Instruct | 53.23 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen3-235B-Instruct | 53.05 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen2.5-32B-Instruct | 52.85 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen2.5-72B-Instruct | 51.86 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen3-32B | 51.67 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen2.5-32B-Instruct | 51.18 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen3-14B | 50.47 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen3-8B | 50.05 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen3-32B | 49.58 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen3-14B | 47.87 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen3-8B | 47.55 | Not recorded | 69c4bdb20636e5bf4afc7717 | | GPT-3.5-Turbo | 43.67 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen2.5-7B-Instruct | 43.23 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Qwen2.5-7B-Instruct | 41.67 | Not recorded | 69c4bdb20636e5bf4afc7717 | | Gemini-2.5-Flash | 40.85 | Not recorded | 69c4bdb20636e5bf4afc7717 | | GPT-3.5-Turbo | 38.47 | Not recorded | 69c4bdb20636e5bf4afc7717 | ### Gemini-2.5-Pro Average Success Rate: 62.15. Result ID: 69c55dd347f34ce924afbd12. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. ### Gemini-2.5-Pro Average Success Rate: 60.98. Result ID: 69c55dd347f34ce924afbd22. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. ### GPT-4o Average Success Rate: 59.67. Result ID: 69c55dd347f34ce924afbd15. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. ### Gemini-2.5-Flash Average Success Rate: 58.35. Result ID: 69c55dd347f34ce924afbd21. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. ### GPT-4o Average Success Rate: 58.26. Result ID: 69c55dd347f34ce924afbd25. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. ### Qwen3-235B-Instruct Average Success Rate: 58.06. Result ID: 69c55dd347f34ce924afbd26. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Claude-4.0-Sonnet Average Success Rate: 57.96. Result ID: 69c55dd347f34ce924afbd13. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. ### FinTool-Qwen3-14B Average Success Rate: 57.95. Result ID: 69c55dd347f34ce924afbd1f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"FinTool-Tuned (Ours)"}. This record does not assert independent reproduction. ### FinTool-Qwen3-14B Average Success Rate: 57.77. Result ID: 69c55dd347f34ce924afbd2e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"FinTool-Tuned (Ours)"}. This record does not assert independent reproduction. ### FinTool-Qwen3-8B Average Success Rate: 57.64. Result ID: 69c55dd347f34ce924afbd2f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"FinTool-Tuned (Ours)"}. This record does not assert independent reproduction. ### Qwen2.5-14B-Instruct Average Success Rate: 57.23. Result ID: 69c55dd347f34ce924afbd29. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Claude-4.0-Sonnet Average Success Rate: 55.93. Result ID: 69c55dd347f34ce924afbd23. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. ### Qwen2.5-14B-Instruct Average Success Rate: 55.75. Result ID: 69c55dd347f34ce924afbd1a. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### FinTool-Qwen3-8B Average Success Rate: 55.68. Result ID: 69c55dd347f34ce924afbd20. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"FinTool-Tuned (Ours)"}. This record does not assert independent reproduction. ### Deepseek-V3.1-Terminus Average Success Rate: 55.32. Result ID: 69c55dd347f34ce924afbd16. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen2.5-72B-Instruct Average Success Rate: 53.23. Result ID: 69c55dd347f34ce924afbd27. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen3-235B-Instruct Average Success Rate: 53.05. Result ID: 69c55dd347f34ce924afbd17. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen2.5-32B-Instruct Average Success Rate: 52.85. Result ID: 69c55dd347f34ce924afbd28. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen2.5-72B-Instruct Average Success Rate: 51.86. Result ID: 69c55dd347f34ce924afbd18. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen3-32B Average Success Rate: 51.67. Result ID: 69c55dd347f34ce924afbd1c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen2.5-32B-Instruct Average Success Rate: 51.18. Result ID: 69c55dd347f34ce924afbd19. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen3-14B Average Success Rate: 50.47. Result ID: 69c55dd347f34ce924afbd2c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen3-8B Average Success Rate: 50.05. Result ID: 69c55dd347f34ce924afbd2d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen3-32B Average Success Rate: 49.58. Result ID: 69c55dd347f34ce924afbd2b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen3-14B Average Success Rate: 47.87. Result ID: 69c55dd347f34ce924afbd1d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen3-8B Average Success Rate: 47.55. Result ID: 69c55dd347f34ce924afbd1e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### GPT-3.5-Turbo Average Success Rate: 43.67. Result ID: 69c55dd347f34ce924afbd24. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. ### Qwen2.5-7B-Instruct Average Success Rate: 43.23. Result ID: 69c55dd347f34ce924afbd2a. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Function Calling (FC) Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Qwen2.5-7B-Instruct Average Success Rate: 41.67. Result ID: 69c55dd347f34ce924afbd1b. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Open-Source LLMs"}. This record does not assert independent reproduction. ### Gemini-2.5-Flash Average Success Rate: 40.85. Result ID: 69c55dd347f34ce924afbd11. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. ### GPT-3.5-Turbo Average Success Rate: 38.47. Result ID: 69c55dd347f34ce924afbd14. [Reporting paper]() Recorded evaluation configuration: {"Evaluation Mode":"Prompt Mode","Model Category":"Closed-Source LLMs"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.