Salesforce
We track 20 benchmarks that Salesforce made or helped make. Its models are on its lab page.
0 in use 18 with public code 12 with a leaderboard here
Benchmarks
| Name | Kind | Released | Best score | Status |
|---|---|---|---|---|
| MMEB-V2 Multimodal embedding models are scored on 78 tasks over images, videos and visual documents. V2 adds visual document retrieval, video retrieval, temporal grounding, video classification and video QA. | General visual questions | May 2025 | | Fading |
| MCP-Universe Agents complete tasks through 11 real MCP servers across six domains, such as navigation, repositories, finance, and browsing. Scored by execution-based success rate. | Apps and MCP servers | Aug 2025 | | Fading |
| CRM-Agent-Bench Agents handle customer and business requests in a simulated enterprise. Executable checks verify tool arguments, action order, and which records changed across 555 tasks. | Running a business | Oct 2026 | | Not in cards |
| EGI-BENCH Agents work across eight enterprise evaluations. The suite combines their results into six scores covering activities such as planning, retrieving evidence, and using tools. | Simulated companies | Oct 2026 | 65.6 | Not in cards |
| LLM-AggreFact Models judge whether claims are supported by a grounding document across 11 pooled datasets, such as RAGTruth and ExpertQA. Scored by balanced accuracy. | Faithful to sources | Apr 2024 | 75.9% | Not in cards |
| CRMArena Agents do nine customer-service CRM tasks for service agent, analyst and manager personas inside a real Salesforce org. The org holds 16 linked object types with seeded hidden patterns. | Customer service | Nov 2024 | 64.3% | Not in cards |
| CRMArena-Pro Agents do 19 expert-checked sales, service and quoting tasks in B2B and B2C Salesforce orgs, in single-turn and persona-driven multi-turn settings. It also tests whether agents keep confidential data private. | Customer service | May 2025 | | Not in cards |
| UserBench Agents help simulated users who start with vague goals and reveal preferences bit by bit. Agents must ask clarifying questions and use tools to reach choices that fit the user. | General assistants | Jul 2025 | 0.61 | Not in cards |
| LiveResearchBench Deep research agents write cited reports for user-centric questions that need current information from many live web sources. Rubric-based judges score the reports. | Deep research | Oct 2025 | 73.7 | Not in cards |
| SCUBA Agents work in a Salesforce sandbox as administrators, sales representatives, or service agents. Evaluators check milestones such as creating accounts and routing support cases. | Web browsing | Sep 2025 | 53.9% | Not in cards |
| AgentScript-Eval Models write, debug, or optimize AgentScript definitions for Salesforce agents. A task passes when the program compiles and passes every execution test. | Function-level problems | Nov 2026 | 37% | Not in cards |
| Summary of a Haystack Models read a synthetic haystack of conversation or news documents where specific insights repeat, then write a query-focused summary that cites its sources. An automatic evaluation scores insight coverage and citation accuracy. | Reasoning over long documents | Jul 2024 | 42.5% | Not in cards |
| GIFT-Eval-CTX Models forecast a time series using past observations and information about future changes. | Real-world events | Oct 2026 | 0.57 | Not in cards |
| MMEB Multimodal embedding models are scored on classification, visual question answering, multimodal retrieval and visual grounding, with 16 evaluation datasets in and out of distribution. | General visual questions | Oct 2024 | | Not in cards |
| ContextualBench CRM agents answer enterprise questions against contextual business data and receive accuracy, cost, speed, trust and safety measurements. | Questions that need search | Sep 2024 | Not in cards | |
| ConvoMem Models answer questions about earlier conversations, including changed facts and user preferences. The dataset contains 75,336 items across 100 fictional users. | Finding facts in long inputs | Nov 2025 | Not in cards | |
| GIFT-Eval Models forecast future values from historical time series across multiple domains and forecast horizons. | Real-world events | Oct 2024 | Not in cards | |
| LLM Economist Models set tax rates or choose working hours in a simulated economy. Researchers compare the social welfare achieved under different tax policies. | Running a business | Jul 2025 | Not in cards | |
| ReasoningJudgeBench Judge models rate reasoning outputs to measure evaluator reliability and agreement across difficult tasks. | AI research tasks | Jan 2026 | Not in cards | |
| WikiSQL Models turn natural language questions into SQL queries over single Wikipedia tables. The dataset has 80,654 hand-annotated questions over 24,241 tables. Scored by execution accuracy and logical form accuracy. | Data science and SQL | Aug 2017 | Not in cards |