UC BerkeleyWe track 46 benchmarks and 3 arenas that UC Berkeley made or helped make. Its models are on its lab page.
Top five in each of its 2 arenas with three or more rated models, the most rated first. We don't have a leaderboard yet for one other, listed below.
Multi-player conquest game with private non-binding negotiation and secret asymmetric objectives. LM agents and humans are ranked by Plackett-Luce player strength from their game results.
Gemini 3.1 Pro 0.74 Gemini 3.1 Flash-Lite -0.42 Official leaderboard
| Name | Kind | Released | Best score | Status |
|---|---|---|---|---|
| MMLU 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy. | Exam questions | Sep 2020 |  92.3% | Fading |
| MATH 12,500 competition math problems with step-by-step solutions. Final answer scored by exact match. | Competition problems with answers | Mar 2021 |  98.2% | Fading |
| LiveCodeBench v6 Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. | Competitive programming | Apr 2025 |  93.5% | In use |
| LiveCodeBench Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. | Competitive programming | Mar 2024 | 93.5% | In use |
| LiveCodeBench v5 Competitive programming problems collected after model cutoffs from LeetCode, AtCoder, Codeforces. Scored pass@1 on hidden tests. | Competitive programming | Jan 2025 |  81.2% | Fading |
| Berkeley Function Calling Leaderboard (BFCL) Model emits function calls for single, parallel, multi-turn and agentic scenarios. Scored by AST and execution checks. | Function calling | Feb 2024 |  88.5% | In use |
| BFCL V4 Models call functions and tools, including multi-turn agentic tasks with web search and memory. Checked against expected calls and outcomes. | Function calling | Jul 2025 |  75% | In use |
| CyberGym Large-scale benchmark of real open-source vulnerabilities. Agent must reproduce each known vulnerability with a proof-of-concept input. | Finding and exploiting real bugs | Jun 2025 | 99.4% | In use |
| Agents' Last Exam 1,500+ expert-sourced tasks from real projects across 55 occupations in three difficulty tiers. Objective verifiers grade deliverables. 5-hour cap. | Tasks from real jobs | Jun 2026 | 38.2% | In use |
| StrongREJECT Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is. | Jailbreaks and prompt injection | Feb 2024 |  98.6% | In use |
| DynaMath Models answer visual math questions generated from 501 seed programs that vary values and graphs. Scored by average and worst-case accuracy across variants. | Visual math and diagrams | Oct 2024 | 88% | Fading |
| ExploitGym 869 real-vulnerability exploit-development tasks (userspace, V8, Linux kernel). Success means capturing a flag through the intended vulnerability. | Finding and exploiting real bugs | May 2026 | 42.4% | In use |
| EgoSchema Models answer five-way multiple-choice questions about three-minute egocentric video clips from Ego4D. Scored by accuracy on 5,031 questions. | Long videos | Aug 2023 | 76.2% | Dropped |
| MMVP Models answer 300 questions about 150 image pairs that CLIP embeds as near duplicates. A pair counts only when both answers are right. | Spatial reasoning and counting | Jan 2024 | 87% | Fading |
| Gorilla APIBench Models write calls to machine learning APIs from Hugging Face, TorchHub and TensorHub. Scored by AST match. | Function calling | May 2023 | 35.3% | System cards only |
| Agents' Last Exam v1.1 Agents complete 122 public long-horizon professional workflow tasks across 55 sub-industries. Deterministic graders check outcomes against hidden references. | Tasks from real jobs | Oct 2026 | 45.9% | In use |
| APPS Models write Python programs for 10,000 coding problems that range from introductory to competition level. Generated code runs against test cases. | Competitive programming | May 2021 | 70.2% | Dropped |
| OMEGA Models solve generated competition-style math problems that test exploratory, compositional and transformative generalization beyond training patterns. Answers are checked automatically. | Competition problems with answers | Jun 2025 |  53.4% | Fading |
| JudgeBench Models act as judges and pick the correct response from hard pairs on knowledge, reasoning, math and coding. The label comes from objective correctness, not human preference. | Overall chat quality | Oct 2024 | 80.9% | Dropped |
| SimplerEnv Real-to-sim replicas of Google Robot and WidowX/Bridge setups. Policies scored by success rate on pick, move, drawer and Bridge tasks. | Simulated robots | May 2024 | 63% | Fading |
| MMLU-stem Models answer the STEM subjects of MMLU, such as physics, chemistry, mathematics, and computer science, as multiple choice. Scoring is accuracy. | Exam questions | Sep 2020 |  83.6% | Dropped |
| RoboDojo Generalist manipulation policies run 42 Isaac Sim tasks and 18 remote real-robot tasks. Scored by task score and success rate. | Simulated robots | Jul 2026 | 29% | In use |
| CyberGym (final-submission metric) Agents reproduce the same 1,507 real vulnerabilities as CyberGym, but a task counts as solved only if the one proof-of-concept the agent picks as final triggers the bug. | Finding and exploiting real bugs | Jul 2026 |  86.3% | In use |
| DiverseVul Models classify C and C++ functions from real vulnerability fixing commits as vulnerable or not. | Finding and exploiting real bugs | Apr 2023 | 57% | System cards only |
| MMLU Anatomy Models answer MMLU multiple-choice questions from the Anatomy subject. Scoring is accuracy. | Medicine | Sep 2020 | 89% | System cards only |
| MMLU Clinical Knowledge Models answer MMLU multiple-choice questions from the Clinical Knowledge subject. Scoring is accuracy. | Medicine | Sep 2020 | 92% | System cards only |
| MMLU College Biology Models answer MMLU multiple-choice questions from the College Biology subject. Scoring is accuracy. | Exam questions | Sep 2020 | 95% | System cards only |
| MMLU College Medicine Models answer MMLU multiple-choice questions from the College Medicine subject. Scoring is accuracy. | Medicine | Sep 2020 | 84% | System cards only |
| MMLU Medical Genetics Models answer MMLU multiple-choice questions from the Medical Genetics subject. Scoring is accuracy. | Medicine | Sep 2020 | 96% | System cards only |
| MMLU Professional Medicine Models answer MMLU multiple-choice questions from the Professional Medicine subject. Scoring is accuracy. | Medicine | Sep 2020 | 94% | System cards only |
| BaxBench Models write backend servers from a specification. Each solution is checked for correctness by tests and for security by end to end exploits. | Secure code and review | Feb 2025 | 68.6% | Not in cards |
| Data Agent Benchmark Agents answer business questions over data split across several database systems, with duplicated and inconsistent records and key facts buried in text. The 104 queries cover 17 datasets and four database systems, based on a study of companies building data agents in six industries. Scores are pass@1 against known answers. | Data science and SQL | Mar 2026 | 99.2% | Not in cards |
| CaP-X (CaP-Bench) Coding agents write perception and control code for 100+ sim manipulation tasks across 8 abstraction and interaction tiers. Scored by task success. | Simulated robots | Mar 2026 | 32.3% | Not in cards |
| Embodied First Exam Coding agents answer frozen robot tasks from many benchmark suites with a program. Each program is verified in a separate sandbox, run on Harbor. | Simulated robots | Oct 2026 | 49.6% | Not in cards |
| Continual Learning Bench AI systems work through sequences of related task instances in six domains and should get better as they go. A gain score subtracts each system's own stateless baseline to isolate learning. | Learning from the context | Jun 2026 | 0.20 | Not in cards |
| CyberGym-E2E Agents get only source code and must find a real vulnerability, write a proof-of-concept input and a patch that keeps functionality. Covers 920 vulnerabilities in 139 open-source projects. | Finding and exploiting real bugs | Jun 2026 | 89% | Not in cards |
| CyberGym-E2E (615 tasks) Agents discover and demonstrate vulnerabilities on the older 615-task cohort. The headline is cumulative stage S3 success. | Finding and exploiting real bugs | Jun 2026 | 22.6% | Not in cards |
| DS-1000 Models write Python for 1,000 data science problems from StackOverflow across seven libraries such as NumPy and Pandas. Test cases and API constraints check each answer. | Data science and SQL | Nov 2022 | 53.9% | Not in cards |
| AutoEval Autonomous real-robot evaluation: WidowX arms run five Bridge tasks with learned success detectors and resets. 50 rollouts per task. | Real robots | Mar 2025 | Not in cards | |
| CUAD Models review commercial contracts and highlight the clauses a lawyer would flag, across 41 categories such as governing law, non-compete and change of control. Legal experts from The Atticus Project wrote more than 13,000 labels over 510 contracts. Scores measure how well the extracted spans match the expert labels. | Law | Mar 2021 | Not in cards | |
| ETHICS Models predict common moral judgments about text scenarios covering justice, well-being, duties, virtues and commonsense morality. | Deception and alignment | Aug 2020 | Not in cards | |
| GaP: Graph-as-Policy Multi-agent harness that compiles language tasks into typed robot skill graphs. Tested on 8 variational-automation tasks (4 sim, 4 real) by success rate. | Simulated robots | Jul 2026 | Not in cards | |
| HumanoidBench Policies control a simulated Unitree H1 humanoid with dexterous hands on 15 whole-body manipulation and 12 locomotion tasks. Scored by task return or success. | Simulated robots | Mar 2024 | Not in cards | |
| MACHIAVELLI Agents play 134 text adventure games with over half a million scenarios. The score is normalized game reward. Annotations also score power-seeking, disutility and ethical violations. | Deception and alignment | Apr 2023 | Not in cards | |
| MMLU Medical Models answer the multiple-choice questions from the medical subjects of MMLU, such as anatomy, clinical knowledge and professional medicine. | Medicine | Sep 2020 | Not in cards | |
| OCBench 28 simulated manipulation tasks with human-like scripted demonstrators for controlled BC and RL studies. Policies scored by task success rate. | Simulated robots | Oct 2026 | Not in cards |
| Name | Kind | Released | Best score | Status |
|---|---|---|---|---|
| Agent Arena (Gorilla) Users give a task, two agents built from different models, tools and frameworks run it, and users vote. A Bradley-Terry extension rates models, tools and frameworks. | General assistants | Oct 2024 | Not in cards |