// category · 17 tools · 2 open source
LLM Leaderboards
Public rankings that compare models on preference, accuracy, speed or price. Read the methodology: each leaderboard measures something different.
Leaderboards rank AI models by combining benchmark results, human votes or private test sets into a single table. They are where most people first see claims such as "the best model right now," so this category lists the main boards and what each one actually measures.
They fall into three types.
Aggregated benchmark indexes. Artificial Analysis runs models through the same suite of evaluations under identical settings and combines them into its Intelligence Index, alongside speed, latency and price per token. The Epoch AI Benchmarking Hub tracks results across many benchmarks over time.
Human preference arenas. Arena, formerly LMArena, asks people to compare two anonymous answers and vote, then turns the votes into ratings, with a style control adjustment for longer or heavily formatted answers. That captures what users prefer, which is not the same as accuracy on hard questions.
Private or specialist boards. Scale's SEAL leaderboards use private expert built datasets, which are hard to train on but impossible for outsiders to inspect. Specialist boards cover narrower skills: the Berkeley Function Calling Leaderboard for tool use, Aider's leaderboards for code editing, MTEB for embedding models, and the Vectara hallucination leaderboard for how often summaries stay faithful to the source.
How to read any leaderboard:
- Check the date. Rankings change with every major release.
- Check the method. A benchmark index, a vote based rating and a private test answer different questions.
- Look at the spread. If confidence intervals overlap, models are effectively tied.
- Match it to your task. A model that tops a chat arena may not lead on code or retrieval.
- Ask who runs it. Some operators also sell services to model developers.
No leaderboard replaces testing on your own data. Use boards to build a shortlist, then run your own evaluation set.
Our LLM leaderboards compared guide covers each board in detail, and LLM benchmarks explained describes the tests underneath them. For a single benchmark that many boards include, see the HLE Diamond guide.
// guides
Guides to LLM leaderboards
LLM Leaderboards Compared: Which Ones to Trust
Every major LLM leaderboard compared: how Arena, Artificial Analysis, LiveBench, SEAL, Epoch AI and Vellum score models, and where each one can mislead you.
Which LLM Benchmark Measures Tool Calling?
Which LLM benchmark measures tool calling? BFCL V4 is the public standard for function calls; Tau bench and ToolBench cover dialogue and large API inventories.
LiveBench vs Artificial Analysis: Key Differences
LiveBench vs Artificial Analysis: how the two AI leaderboards differ on tasks, grading, weighting and cost, why their rankings disagree, and when to trust each.
LLM Benchmarks Explained: What the Scores Mean
LLM benchmarks explained: what MMLU, GPQA Diamond, Humanity's Last Exam, SWE-bench and ARC-AGI measure, why scores saturate, and how to spot contamination.
Gemini 4 Neon Benchmarks: Which Claims Hold Up
Gemini 4 Neon benchmarks checked against Artificial Analysis, Vals, Arena, Terminal-Bench, Zapier and Cognition boards: which launch claims match and which do not.
Gemini 4 Hydrogen Benchmarks: What Holds Up
Gemini 4 Hydrogen benchmarks checked against Artificial Analysis, Vals, Terminal-Bench, Snorkel and Arena boards: which launch scores hold up and which do not.
GDPx Benchmark Explained: Tasks, Scores, Limits
The GDPx benchmark from SemiAnalysis tests AI agents on 206 hard tasks in 11 domains. How it is scored, what the first results show, and where it misleads.
Agent Arena Index: How Arena Ranks AI Agents
The Agent Arena Index explained: the four signals behind Arena's agent leaderboard, how the score is weighted and anchored, and how to read the November 2026 board.
SO-Arena Benchmark: How System One Models Rank
SO-Arena tests System One decision models on 2,919 tasks in 11 domains. Who leads, why the top scorer is only seventh on reliability, and how to read it.
ARC-AGI vs Humanity's Last Exam: Key Differences
ARC-AGI vs Humanity's Last Exam: one tests how fast a model learns a new puzzle, the other tests expert knowledge. Scores, grading, cost and which to trust.
Jev Decision Index: How Decision Models Rank
The Jev Decision Index ranks 275 decision models on private tests. How the Full score works, who leads, why Jev sits 55th, and how it differs from AA's index.
Harness Leaderboard: How the Harness Moves Scores
The Harness Leaderboard runs 9 coding agent harnesses on the same open models. What v0.1.3 shows, who ties for first, what it costs, and the caveats.
Terminal-Bench Explained: 4.0, 4.1, Scores
Terminal-Bench explained: how the 66 task agent shell benchmark works, what changed in 4.0 and 4.1, and how to read official, Vals and Artificial Analysis scores.
Fireworks Specialized Intelligence Index Explained
The Fireworks Specialized Intelligence Index tracks 66 real work AI benchmarks in 14 domains. How it scores models, what changed on Oct 28, and how to read it.
Artificial Analysis Intelligence Index v5 Explained
Artificial Analysis Intelligence Index v5: the nine evaluations, new 25% weights, what was dropped, why old scores no longer compare, and how to read a v5 score.
FrontierCode Benchmark Explained: How It Scores
The FrontierCode benchmark explained: how Cognition grades whether maintainers would merge AI code, what Main and Diamond mean, and how to read the leaderboards.
MMLU vs GPQA: Key Differences and Which to Use
MMLU vs GPQA explained: broad exam recall versus expert written science questions, why GPQA scores run lower, and which benchmark still means anything in 2026.
SWE-Bench Pro Explained: V1, V2, Scores
SWE-Bench Pro explained: how Scale AI's coding benchmark differs from Verified, what V2 changed, and the November 2026 full, HARD-51 and private scores.
HLE Diamond Leaderboard and Scores (November 2026)
HLE Diamond leaderboard for November 2026: Claude Fable 5.2 leads both official tables, 72.4% on Scale's 21 model board and 74.3% in the authors' table.
VIALS-AA: What Scientific Image Scores Mean
VIALS-AA is Artificial Analysis's independent board for Handshake AI's 161 task scientific image test. How pass cubed works, what scores show, and the limits.
Coding Agent Benchmark Rankings (October 2026)
Coding agent benchmark rankings from SWE-Bench Pro, FrontierCode, Terminal-Bench and the Artificial Analysis index, with sourced scores, dates and why they differ.
What Is Faithfulness in RAG Evaluation?
What is faithfulness in RAG evaluation? The share of an answer's claims backed by the retrieved context, and how Ragas, DeepEval, TruLens and HHEM score it.
GDPval Benchmark: 1,320 Real Work Tasks, 44 Jobs
GDPval is OpenAI's AI test of 1,320 real work tasks from 44 occupations. How grading works, how to read the GDPval-AA board, and what it misses.
How to Evaluate LLM Model Performance
How to evaluate LLM model performance before you pick one: build a task test set, fix settings, add error bars, test prompt sensitivity, then weigh cost.