# LLMBenchmark.ai

> LLMBenchmark.ai is an LLM benchmark leaderboard that tracks 356 AI models across 190 public benchmarks, with API prices, updated daily.

Last data refresh: October 20, 2026 (2026-10-20). The dataset is rebuilt from its sources every day.

## What is original here

- Vibe Coding Index: a 0 to 100 score that averages 4 public coding benchmarks, weighted as Aider polyglot pass rate (35%), SWE-bench Verified resolved rate (30%), LiveBench coding score (20%), WebDev Arena Elo (15%). A benchmark a model was never run on is skipped, never estimated. 136 of 356 models have a score.
- Evidence tiers: each score is labelled by how many benchmarks stand behind it. 2 models have three or more, 64 models have two, 70 models have one. Models with none are listed without a score.
- Cost per solved task: the cost of one SWE-bench Verified attempt divided by the share of tasks the model resolves. 84 models have both numbers. A cheaper price per token can still mean a dearer working answer.
- Daily price snapshots, so a price change shows with its date. Output prices run from $0.0112 to $600 per million tokens, median $1.8.

## Sections

- [Ranking](https://llmbenchmark.ai/rankings/vibe-coding): every scored model by index, with the benchmarks behind each score; a filter keeps scores from two or more. The flagship.
- [Rankings beyond coding](https://llmbenchmark.ai/rankings): reasoning (GPQA Diamond), math (AIME), knowledge (MMLU Pro) and general skill (Vals Index), each one public board shown as published and kept out of the coding index.
- [Best for lists](https://llmbenchmark.ai/best): 88 lists generated from the data, such as the cheapest model per solved task, the best open-weight model for coding and the best model from each maker.
- [New and trending](https://llmbenchmark.ai/trending): models released in the last 30 days, and open-weight models by Hugging Face trending score and 30-day downloads. Closed models publish no usage figures and are not ranked on popularity.
- [Embed](https://llmbenchmark.ai/embed): charts and comparison tables, licensed CC BY 4.0, free to embed with a visible source link.
- [API and MCP server](https://llmbenchmark.ai/api): free, no key, 500 requests per IP per day. Ask which model fits a task and get an answer from our data.
- [Model updates](https://llmbenchmark.ai/model-updates): what changed in the catalog, newest first. Also as a feed: [RSS](https://llmbenchmark.ai/rss.xml).
- [Contact](https://llmbenchmark.ai/contact): how to reach the team behind LLMBenchmark.ai.
- [Models](https://llmbenchmark.ai/models): one page per model with price, context window, every benchmark result we hold and cost per solved task. Missing data says so.
- Comparisons: side-by-side pages for models with at least two benchmarks, linked from each model page.
- [Pricing](https://llmbenchmark.ai/pricing): API prices by provider, with a [cost calculator](https://llmbenchmark.ai/pricing/calculator).
- [Cost per task](https://llmbenchmark.ai/cost-per-task): the formula, a worked example and the ranking of models by cost.
- [Vibe coding](https://llmbenchmark.ai/vibe-coding): what the term means, who coined it, with dated sources, and its [tools](https://llmbenchmark.ai/vibe-coding/tools).
- [Methodology](https://llmbenchmark.ai/methodology): weights, score ranges, coverage rules and what the index leaves out.
- [About](https://llmbenchmark.ai/about): who runs LLMBenchmark.ai, where the numbers come from, how often they change and the company behind it.

## Benchmarks

- [Aider polyglot](https://llmbenchmark.ai/benchmarks/aider-polyglot): in the index. Source: Aider polyglot benchmark write-up.
- [SWE-bench Verified](https://llmbenchmark.ai/benchmarks/swe-bench-verified): in the index. Source: SWE-bench: Can Language Models Resolve Real-World GitHub Issues?.
- [SWE-bench Lite](https://llmbenchmark.ai/benchmarks/swe-bench-lite): tracked, not in the index. Source: SWE-bench: Can Language Models Resolve Real-World GitHub Issues?.
- [SWE-bench (full test set)](https://llmbenchmark.ai/benchmarks/swe-bench-full): tracked, not in the index. Source: SWE-bench: Can Language Models Resolve Real-World GitHub Issues?.
- [SWE-bench Multimodal](https://llmbenchmark.ai/benchmarks/swe-bench-multimodal): tracked, not in the index. Source: SWE-bench Multimodal: Do AI Systems Generalize to Visual Software Domains?.
- [SWE-bench Multilingual](https://llmbenchmark.ai/benchmarks/swe-bench-multilingual): tracked, not in the index. Source: SWE-bench Multilingual.
- [LiveBench coding](https://llmbenchmark.ai/benchmarks/livebench-coding): in the index. Source: LiveBench: A Challenging, Contamination-Limited LLM Benchmark.
- [LMArena WebDev](https://llmbenchmark.ai/benchmarks/lmarena-webdev): in the index. Source: Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference.
- [Vals Index](https://llmbenchmark.ai/benchmarks/vals-index): tracked, not in the index. Source: Vals Index.
- [GPQA Diamond](https://llmbenchmark.ai/benchmarks/gpqa): tracked, not in the index. Source: GPQA: A Graduate-Level Google-Proof Q&A Benchmark.
- [AIME](https://llmbenchmark.ai/benchmarks/aime): tracked, not in the index. Source: AIME benchmark, Vals AI.
- [MMLU Pro](https://llmbenchmark.ai/benchmarks/mmlu-pro): tracked, not in the index. Source: MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmark.
- [AA Intelligence Index](https://llmbenchmark.ai/benchmarks/artificial-analysis-intelligence-index): tracked, not in the index. Source: Artificial Analysis methodology.
- [AA Coding Index](https://llmbenchmark.ai/benchmarks/artificial-analysis-coding-index): tracked, not in the index. Source: Artificial Analysis methodology.
- [AA Agentic Index](https://llmbenchmark.ai/benchmarks/artificial-analysis-agentic-index): tracked, not in the index. Source: Artificial Analysis methodology.
- [Design Arena: websites](https://llmbenchmark.ai/benchmarks/design-arena-website): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: UI components](https://llmbenchmark.ai/benchmarks/design-arena-ui): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: game dev](https://llmbenchmark.ai/benchmarks/design-arena-gamedev): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: data viz](https://llmbenchmark.ai/benchmarks/design-arena-dataviz): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: 3D](https://llmbenchmark.ai/benchmarks/design-arena-3d): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: all code categories](https://llmbenchmark.ai/benchmarks/design-arena-code-overall): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: SVG](https://llmbenchmark.ai/benchmarks/design-arena-svg): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: ASCII art](https://llmbenchmark.ai/benchmarks/design-arena-ascii-art): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: agents building web apps](https://llmbenchmark.ai/benchmarks/design-arena-agents-web-apps): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: agents building full-stack apps](https://llmbenchmark.ai/benchmarks/design-arena-agents-full-stack): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: agents building mobile apps](https://llmbenchmark.ai/benchmarks/design-arena-agents-mobile-apps): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: agents building Android apps](https://llmbenchmark.ai/benchmarks/design-arena-agents-android): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: agents building Godot games](https://llmbenchmark.ai/benchmarks/design-arena-agents-godot): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: agents building games](https://llmbenchmark.ai/benchmarks/design-arena-agents-games): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: HTML slides](https://llmbenchmark.ai/benchmarks/design-arena-agents-html-slides): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: PowerPoint slides](https://llmbenchmark.ai/benchmarks/design-arena-agents-pptx-slides): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: spreadsheets](https://llmbenchmark.ai/benchmarks/design-arena-agents-spreadsheets): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: images](https://llmbenchmark.ai/benchmarks/design-arena-images): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: graphic design](https://llmbenchmark.ai/benchmarks/design-arena-graphic-design): tracked, not in the index. Source: Design Arena leaderboard.
- [Design Arena: logos](https://llmbenchmark.ai/benchmarks/design-arena-logos): tracked, not in the index. Source: Design Arena leaderboard.
- [LiveBench reasoning](https://llmbenchmark.ai/benchmarks/livebench-reasoning): tracked, not in the index. Source: LiveBench.
- [LiveBench data analysis](https://llmbenchmark.ai/benchmarks/livebench-data-analysis): tracked, not in the index. Source: LiveBench.
- [LiveBench research](https://llmbenchmark.ai/benchmarks/livebench-research): tracked, not in the index. Source: LiveBench.
- [LiveBench browser](https://llmbenchmark.ai/benchmarks/livebench-browser): tracked, not in the index. Source: LiveBench.
- [LiveBench long agentic](https://llmbenchmark.ai/benchmarks/livebench-long-agentic): tracked, not in the index. Source: LiveBench.
- [LiveBench design](https://llmbenchmark.ai/benchmarks/livebench-design): tracked, not in the index. Source: LiveBench.
- [LiveCodeBench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-livecodebench): tracked, not in the index. Source: LiveCodeBench (Vals AI).
- [SWE-bench Verified (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-swe-bench): tracked, not in the index. Source: SWE-bench Verified (Vals AI).
- [Terminal-Bench 4 (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-terminal-bench-4): tracked, not in the index. Source: Terminal-Bench 4 (Vals AI).
- [Vibe Code Bench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-vibe-code): tracked, not in the index. Source: Vibe Code Bench (Vals AI).
- [Code migration (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-code-migration): tracked, not in the index. Source: Code migration (Vals AI).
- [ProgramBench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-programbench): tracked, not in the index. Source: ProgramBench (Vals AI).
- [SREBench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-srebench): tracked, not in the index. Source: SREBench (Vals AI).
- [IOI (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-ioi): tracked, not in the index. Source: IOI (Vals AI).
- [Terminal-Bench 2.1 (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-terminal-bench-2-1): tracked, not in the index. Source: Terminal-Bench 2.1 (Vals AI).
- [SWE-bench Multimodal (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-swe-bench-multimodal): tracked, not in the index. Source: SWE-bench Multimodal (Vals AI).
- [Vibe Code Bench 1-100 (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-vibe-code-1-100): tracked, not in the index. Source: Vibe Code Bench 1-100 (Vals AI).
- [SkillsBench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-skillsbench): tracked, not in the index. Source: SkillsBench (Vals AI).
- [CyberBench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-cyberbench): tracked, not in the index. Source: CyberBench (Vals AI).
- [Finance Agent (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-finance-agent): tracked, not in the index. Source: Finance Agent (Vals AI).
- [Tax Agent Bench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-tax-agent): tracked, not in the index. Source: Tax Agent Bench (Vals AI).
- [Legal Research Bench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-legal-research): tracked, not in the index. Source: Legal Research Bench (Vals AI).
- [Harvey's Legal Agent Benchmark (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-harvey-legal-agent): tracked, not in the index. Source: Harvey's Legal Agent Benchmark (Vals AI).
- [Excel Modeling Benchmark (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-excel-modeling): tracked, not in the index. Source: Excel Modeling Benchmark (Vals AI).
- [BioMysteryBench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-biomysterybench): tracked, not in the index. Source: BioMysteryBench (Vals AI).
- [Terminal-Bench-Science (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-terminal-bench-science): tracked, not in the index. Source: Terminal-Bench-Science (Vals AI).
- [Vals Science Index](https://llmbenchmark.ai/benchmarks/vals-science-index): tracked, not in the index. Source: Vals Science Index.
- [Time Horizon Index (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-time-horizon-index): tracked, not in the index. Source: Time Horizon Index (Vals AI).
- [CUA-bench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-cua-bench): tracked, not in the index. Source: CUA-bench (Vals AI).
- [Robo-bench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-robo-bench): tracked, not in the index. Source: Robo-bench (Vals AI).
- [Public Benefits Bench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-public-benefits): tracked, not in the index. Source: Public Benefits Bench (Vals AI).
- [MysteryMechanism (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-mysterymechanism): tracked, not in the index. Source: MysteryMechanism (Vals AI).
- [MMMU Pro (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-mmmu-pro): tracked, not in the index. Source: MMMU Pro (Vals AI).
- [MATH 500 (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-math-500): tracked, not in the index. Source: MATH 500 (Vals AI).
- [MGSM (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-mgsm): tracked, not in the index. Source: MGSM (Vals AI).
- [SAGE (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-sage): tracked, not in the index. Source: SAGE (Vals AI).
- [LegalBench (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-legalbench): tracked, not in the index. Source: LegalBench (Vals AI).
- [CaseLaw (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-caselaw): tracked, not in the index. Source: CaseLaw (Vals AI).
- [CorpFin (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-corpfin): tracked, not in the index. Source: CorpFin (Vals AI).
- [TaxEval (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-taxeval): tracked, not in the index. Source: TaxEval (Vals AI).
- [MortgageTax (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-mortgagetax): tracked, not in the index. Source: MortgageTax (Vals AI).
- [MedQA (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-medqa): tracked, not in the index. Source: MedQA (Vals AI).
- [MedCode (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-medcode): tracked, not in the index. Source: MedCode (Vals AI).
- [MedScribe (Vals AI)](https://llmbenchmark.ai/benchmarks/vals-medscribe): tracked, not in the index. Source: MedScribe (Vals AI).
- [Vals Multimodal Index](https://llmbenchmark.ai/benchmarks/vals-multimodal-index): tracked, not in the index. Source: Vals Multimodal Index.
- [AA Math Index](https://llmbenchmark.ai/benchmarks/aa-math-index): tracked, not in the index. Source: Artificial Analysis.
- [GPQA Diamond (Artificial Analysis run)](https://llmbenchmark.ai/benchmarks/aa-gpqa): tracked, not in the index. Source: Artificial Analysis.
- [Humanity's Last Exam (Artificial Analysis run)](https://llmbenchmark.ai/benchmarks/aa-hle): tracked, not in the index. Source: Artificial Analysis.
- [MMLU Pro (Artificial Analysis run)](https://llmbenchmark.ai/benchmarks/aa-mmlu-pro): tracked, not in the index. Source: Artificial Analysis.
- [LiveCodeBench (Artificial Analysis run)](https://llmbenchmark.ai/benchmarks/aa-livecodebench): tracked, not in the index. Source: Artificial Analysis.
- [SciCode](https://llmbenchmark.ai/benchmarks/aa-scicode): tracked, not in the index. Source: Artificial Analysis.
- [Terminal-Bench Hard (Artificial Analysis run)](https://llmbenchmark.ai/benchmarks/aa-terminal-bench-hard): tracked, not in the index. Source: Artificial Analysis.
- [IFBench](https://llmbenchmark.ai/benchmarks/aa-ifbench): tracked, not in the index. Source: Artificial Analysis.
- [Long context reasoning (AA-LCR)](https://llmbenchmark.ai/benchmarks/aa-long-context-reasoning): tracked, not in the index. Source: Artificial Analysis.
- [τ²-Bench Telecom](https://llmbenchmark.ai/benchmarks/aa-tau2): tracked, not in the index. Source: Artificial Analysis.
- [AIME 2025 (Artificial Analysis run)](https://llmbenchmark.ai/benchmarks/aa-aime-2025): tracked, not in the index. Source: Artificial Analysis.
- [Terminal-Bench](https://llmbenchmark.ai/benchmarks/epoch-terminal-bench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [DeepSWE](https://llmbenchmark.ai/benchmarks/epoch-deepswe): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [FrontierCode](https://llmbenchmark.ai/benchmarks/epoch-frontiercode): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [FrontierSWE](https://llmbenchmark.ai/benchmarks/epoch-frontierswe): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [GSO-Bench](https://llmbenchmark.ai/benchmarks/epoch-gso): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [WeirdML](https://llmbenchmark.ai/benchmarks/epoch-weirdml): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [MirrorCode](https://llmbenchmark.ai/benchmarks/epoch-mirrorcode): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [APEX-Agents](https://llmbenchmark.ai/benchmarks/epoch-apex-agents): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [METR time horizons](https://llmbenchmark.ai/benchmarks/epoch-metr-time-horizons): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [Remote Labor Index](https://llmbenchmark.ai/benchmarks/epoch-remote-labor-index): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [GDPval](https://llmbenchmark.ai/benchmarks/epoch-gdpval): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [OSWorld 2.0](https://llmbenchmark.ai/benchmarks/epoch-osworld-2): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [BALROG](https://llmbenchmark.ai/benchmarks/epoch-balrog): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [DeepResearch Bench](https://llmbenchmark.ai/benchmarks/epoch-deepresearch-bench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [PostTrainBench](https://llmbenchmark.ai/benchmarks/epoch-posttrainbench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [GPQA Diamond (Epoch AI run)](https://llmbenchmark.ai/benchmarks/epoch-gpqa-diamond): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [ARC-AGI](https://llmbenchmark.ai/benchmarks/epoch-arc-agi): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [ARC-AGI-2](https://llmbenchmark.ai/benchmarks/epoch-arc-agi-2): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [Humanity's Last Exam](https://llmbenchmark.ai/benchmarks/epoch-hle): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [SimpleBench](https://llmbenchmark.ai/benchmarks/epoch-simplebench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [Chess puzzles](https://llmbenchmark.ai/benchmarks/epoch-chess-puzzles): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [Mystery game puzzles](https://llmbenchmark.ai/benchmarks/epoch-mystery-game-puzzles): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [EBR-bench](https://llmbenchmark.ai/benchmarks/epoch-ebr-bench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [OTIS Mock AIME](https://llmbenchmark.ai/benchmarks/epoch-otis-mock-aime): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [FrontierMath, tiers 1 to 3](https://llmbenchmark.ai/benchmarks/epoch-frontiermath): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [FrontierMath, tier 4](https://llmbenchmark.ai/benchmarks/epoch-frontiermath-tier-4): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [ProofBench](https://llmbenchmark.ai/benchmarks/epoch-proofbench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [MATH level 5](https://llmbenchmark.ai/benchmarks/epoch-math-level-5): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [GSM8K](https://llmbenchmark.ai/benchmarks/epoch-gsm8k): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [SimpleQA Verified](https://llmbenchmark.ai/benchmarks/epoch-simpleqa-verified): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [MMLU](https://llmbenchmark.ai/benchmarks/epoch-mmlu): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [GeoBench](https://llmbenchmark.ai/benchmarks/epoch-geobench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [DTBench](https://llmbenchmark.ai/benchmarks/epoch-dtbench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [LMCA](https://llmbenchmark.ai/benchmarks/epoch-lmca): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [Fiction.LiveBench](https://llmbenchmark.ai/benchmarks/epoch-fiction-livebench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [Creative writing (Lech Mazur)](https://llmbenchmark.ai/benchmarks/epoch-lech-mazur-writing): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [CL-bench](https://llmbenchmark.ai/benchmarks/epoch-cl-bench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [Surface Evolver Bench](https://llmbenchmark.ai/benchmarks/epoch-surface-evolver): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [Furniture assembly](https://llmbenchmark.ai/benchmarks/epoch-furniture-assembly): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [ExploitBench](https://llmbenchmark.ai/benchmarks/epoch-exploitbench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [Cybench](https://llmbenchmark.ai/benchmarks/epoch-cybench): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [TheAgentCompany](https://llmbenchmark.ai/benchmarks/epoch-the-agent-company): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [OSWorld](https://llmbenchmark.ai/benchmarks/epoch-osworld): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [SpireClimb](https://llmbenchmark.ai/benchmarks/epoch-spireclimb): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [VPCT](https://llmbenchmark.ai/benchmarks/epoch-vpct): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [BIG-Bench Hard](https://llmbenchmark.ai/benchmarks/epoch-bbh): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [AI2 Reasoning Challenge](https://llmbenchmark.ai/benchmarks/epoch-arc-ai2): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [HellaSwag](https://llmbenchmark.ai/benchmarks/epoch-hellaswag): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [WinoGrande](https://llmbenchmark.ai/benchmarks/epoch-winogrande): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [PIQA](https://llmbenchmark.ai/benchmarks/epoch-piqa): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [CL-bench Life](https://llmbenchmark.ai/benchmarks/epoch-cl-bench-life): tracked, not in the index. Source: Epoch AI benchmarking hub.
- [ALE-Bench](https://llmbenchmark.ai/benchmarks/ale-bench): tracked, not in the index. Source: ALE-Bench leaderboard, Sakana AI.
- [Vending-Bench 2](https://llmbenchmark.ai/benchmarks/epoch-vending-bench-2): tracked, not in the index. Source: Vending-Bench 2, Andon Labs.
- [Epoch Capabilities Index](https://llmbenchmark.ai/benchmarks/epoch-capabilities-index): tracked, not in the index. Source: Epoch Capabilities Index.
- [LMArena Text](https://llmbenchmark.ai/benchmarks/lmarena-text): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: coding](https://llmbenchmark.ai/benchmarks/lmarena-text-coding): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: hard prompts](https://llmbenchmark.ai/benchmarks/lmarena-hard-prompts): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: math](https://llmbenchmark.ai/benchmarks/lmarena-math): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: instruction following](https://llmbenchmark.ai/benchmarks/lmarena-instruction-following): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: creative writing](https://llmbenchmark.ai/benchmarks/lmarena-creative-writing): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision](https://llmbenchmark.ai/benchmarks/lmarena-vision): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: English](https://llmbenchmark.ai/benchmarks/lmarena-text-english): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: non-English](https://llmbenchmark.ai/benchmarks/lmarena-text-non-english): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: Chinese](https://llmbenchmark.ai/benchmarks/lmarena-text-chinese): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: German](https://llmbenchmark.ai/benchmarks/lmarena-text-german): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: Japanese](https://llmbenchmark.ai/benchmarks/lmarena-text-japanese): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: Korean](https://llmbenchmark.ai/benchmarks/lmarena-text-korean): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: French](https://llmbenchmark.ai/benchmarks/lmarena-text-french): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: Spanish](https://llmbenchmark.ai/benchmarks/lmarena-text-spanish): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: Russian](https://llmbenchmark.ai/benchmarks/lmarena-text-russian): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: Polish](https://llmbenchmark.ai/benchmarks/lmarena-text-polish): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: multi-turn](https://llmbenchmark.ai/benchmarks/lmarena-text-multi-turn): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: longer queries](https://llmbenchmark.ai/benchmarks/lmarena-text-longer-query): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: expert](https://llmbenchmark.ai/benchmarks/lmarena-text-expert): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: hard prompts in English](https://llmbenchmark.ai/benchmarks/lmarena-hard-prompts-english): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: software and IT](https://llmbenchmark.ai/benchmarks/lmarena-text-software): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: mathematical work](https://llmbenchmark.ai/benchmarks/lmarena-text-mathematical): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: science](https://llmbenchmark.ai/benchmarks/lmarena-text-science): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: medicine and healthcare](https://llmbenchmark.ai/benchmarks/lmarena-text-medicine): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: legal and government](https://llmbenchmark.ai/benchmarks/lmarena-text-legal): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: business and finance](https://llmbenchmark.ai/benchmarks/lmarena-text-business): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: writing and language](https://llmbenchmark.ai/benchmarks/lmarena-text-writing): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text: entertainment and media](https://llmbenchmark.ai/benchmarks/lmarena-text-media): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena WebDev: React](https://llmbenchmark.ai/benchmarks/lmarena-webdev-react): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena WebDev: HTML](https://llmbenchmark.ai/benchmarks/lmarena-webdev-html): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena WebDev: from an image](https://llmbenchmark.ai/benchmarks/lmarena-webdev-from-image): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision: English](https://llmbenchmark.ai/benchmarks/lmarena-vision-english): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision: Chinese](https://llmbenchmark.ai/benchmarks/lmarena-vision-chinese): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision: reading text](https://llmbenchmark.ai/benchmarks/lmarena-vision-ocr): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision: diagrams](https://llmbenchmark.ai/benchmarks/lmarena-vision-diagram): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision: homework](https://llmbenchmark.ai/benchmarks/lmarena-vision-homework): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision: humor](https://llmbenchmark.ai/benchmarks/lmarena-vision-humor): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision: captioning](https://llmbenchmark.ai/benchmarks/lmarena-vision-captioning): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision: recognizing things](https://llmbenchmark.ai/benchmarks/lmarena-vision-entities): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Vision: creative writing](https://llmbenchmark.ai/benchmarks/lmarena-vision-creative-writing): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Search](https://llmbenchmark.ai/benchmarks/lmarena-search): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Document](https://llmbenchmark.ai/benchmarks/lmarena-document): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Text to Image](https://llmbenchmark.ai/benchmarks/lmarena-text-to-image): tracked, not in the index. Source: LMArena leaderboard.
- [LMArena Image Edit](https://llmbenchmark.ai/benchmarks/lmarena-image-edit): tracked, not in the index. Source: LMArena leaderboard.

## API and MCP

- MCP server (Streamable HTTP, no key): https://api.llmbenchmark.ai/mcp
- Claude Code: `claude mcp add --transport http llmbenchmark https://api.llmbenchmark.ai/mcp`
- Until api.llmbenchmark.ai points at it, the same server answers at https://vibebench-api.takegroup.work/mcp
- Tools: recommend_model (task text, price, context and open-weights limits in; ranked models out), get_leaderboard, compare_models, get_model, get_prices, list_categories.
- REST twin of recommend_model: https://api.llmbenchmark.ai/v1/recommend?task=refactor+a+large+React+codebase&max_output_price_per_m=5
- Docs: https://llmbenchmark.ai/api
- Limit: 500 requests per IP per UTC day on /mcp and /v1/*, then HTTP 429. Every answer carries data_date and source LLMBenchmark.ai.

## License

- The data is licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Credit "LLMBenchmark.ai" with a link to https://llmbenchmark.ai. Source boards keep their own terms; each score names its source URL and the day it was read.

## Data

- [Full dataset as JSON](https://llmbenchmark.ai/data/dataset.json): models, scores with their source URLs and observation dates, and source metadata. Cite it with the refresh date above.
- [Full text for language models](https://llmbenchmark.ai/llms-full.txt): this brief plus one block per model with its price, context window, rank and every benchmark result, each with its source and the day it was read. Each of those models also has a Markdown page at /models/<slug>.md.

## Sources

- [OpenRouter](https://openrouter.ai/): read October 20, 2026, 356 rows.
- [Artificial Analysis (via OpenRouter)](https://artificialanalysis.ai/): read October 20, 2026, 213 rows.
- [Aider polyglot leaderboard](https://aider.chat/docs/leaderboards/): read October 20, 2026, 22 rows.
- [ALE-Bench (Sakana AI)](https://sakanaai.github.io/ALE-Bench-Leaderboard/): read October 20, 2026, 104 rows.
- [AIME (Vals AI)](https://www.vals.ai/benchmarks/aime): read October 20, 2026, 54 rows.
- [LMArena WebDev](https://lmarena.ai/leaderboard/webdev): read October 20, 2026, 101 rows.
- [SWE-bench (swebench.com)](https://www.swebench.com/): read October 20, 2026, 38 rows.
- [GPQA (Vals AI)](https://www.vals.ai/benchmarks/gpqa): read October 20, 2026, 91 rows.
- [LiveBench](https://livebench.ai/): read October 20, 2026, 43 rows.
- [MMLU Pro (Vals AI)](https://www.vals.ai/benchmarks/mmlu_pro): read October 20, 2026, 92 rows.
- [Vals Index](https://www.vals.ai/benchmarks/vals_index): read October 20, 2026, 20 rows.
- [Artificial Analysis](https://artificialanalysis.ai/): read October 20, 2026, 195 rows.
- [OpenRouter (provider speed)](https://openrouter.ai/): read October 20, 2026, 175 rows.
- [Hugging Face](https://huggingface.co/models): read October 20, 2026, 151 rows.
- [Epoch AI benchmarking hub](https://epoch.ai/benchmarks): read October 20, 2026, 198 rows.
- [Vals AI boards](https://www.vals.ai/benchmarks): read October 20, 2026, 132 rows.
- [LMArena (Hugging Face dataset)](https://lmarena.ai/leaderboard): read October 20, 2026, 169 rows.
- [Design Arena](https://www.designarena.ai/): read October 20, 2026, 141 rows.

## Limits

- No benchmark measures vibe coding directly. The index is an average of coding benchmarks, not a verdict.
- Prices are list prices from OpenRouter, not what any one provider charges directly.