Skip to content

AI Model Evaluation Sources: How to Judge Whether a Model Is Actually Good

Aug 24, 2026 1 min
TL;DR You cannot take model vendors' self-reported scores at face value. This guide covers the most important independent evaluation platforms, domain benchmarks, adoption indicators, and official sources in 2026: what each measures, how to read it, where it is biased, and which figures matter for different use cases.
Table of Contents
  1. General Evaluation Platforms
    1. Artificial Analysis
    2. LMArena (Chatbot Arena)
    3. LiveBench
  2. Domain Benchmarks
    1. SWE-bench / SWE-bench Verified
    2. Aider Polyglot
    3. HumanEval
    4. MMLU-Pro
    5. GPQA Diamond
    6. AIME (American Invitational Mathematics Examination)
    7. Humanity's Last Exam (HLE)
  3. Multimodal Model Evaluation (VLM)
    1. MMMU / MMMU-Pro
    2. MathVista
    3. DocVQA / ChartQA / OCRBench
  4. Image Generation Evaluation
    1. LMArena Text-to-Image Arena
    2. Artificial Analysis Image Arena
    3. Evalytic
    4. Traditional Metrics: FID / CLIP Score
  5. Video Generation Evaluation
    1. Artificial Analysis Video Arena
    2. LMArena Image-to-Video Arena
    3. VBench / VBench-2.0
  6. Speech and Music Evaluation
    1. Artificial Analysis Speech Arena
    2. TTS-Bench
    3. MINT-Bench
    4. Speech Recognition (ASR) Metrics
    5. Music Generation Evaluation
  7. Embedding and Rerank Evaluation
    1. MTEB Leaderboard
    2. BEIR
  8. Market Adoption Indicators
    1. OpenRouter Rankings
    2. Hugging Face Downloads / Likes
    3. Ollama Library
  9. Official Sources
    1. Vendor Model Cards / Blogs
  10. Common Traps
    1. Benchmark Gaming
    2. Leaderboard Saturation
    3. Self-Reported Numbers
    4. Cherry-Picking
    5. Stale Data
  11. Selection Framework: What to Read by Scenario
    1. Agents / Tool Calling
    2. RAG / Search
    3. Cost / High Throughput
    4. General Chat / Customer Service
    5. Reasoning / Math / Science
    6. Image Generation
    7. Video Generation
    8. Text-to-Speech (TTS)
    9. Multimodal Understanding (VLM)
  12. Overall
  13. References

🌏 中文版

Model launches arrive with a wall of benchmark numbers. Which are credible, and which are inflated? This guide breaks down the main evaluation sources in 2026: what each measures, how to interpret it, and where its biases lie.

The goal is not to list every benchmark. It is to help you ask where a number came from, what it represents, and whether to trust it.

General Evaluation Platforms

These three platforms approach the question “how good is the model overall?” from different angles.

Artificial Analysis

artificialanalysis.ai

Measures: quality, price, speed, and latency across text, image, video, and speech. It maintains an Intelligence Index and Coding Agent Index and monitors real API throughput and cost.

How to read it: standardized price and speed comparisons are more valuable than the quality ranking. For API selection, its output tokens/sec and $/1M-token figures are often more useful than one-off local tests because it continuously monitors providers. Its proprietary Intelligence Index is not directly equivalent to other benchmarks.

Biases: primarily covers API models; local models receive less coverage. The Coding Agent Index methodology is less transparent than SWE-bench.

LMArena (Chatbot Arena)

lmarena.ai, formerly lmsys.org.

Measures: human preference. Users chat with two anonymous models, choose the better response, and produce an ELO ranking.

How to read it: ELO answers “which response does an ordinary user prefer?” It is closer to actual user experience than an exam. Differences below 20 points are generally not statistically significant.

Biases: voters favor longer, more formatted answers; style can beat correctness; prompts skew toward ordinary conversation rather than coding or professional work; experienced voters may infer model identity from style. Despite this, Arena is widely cited because static benchmarks are easier to game.

LiveBench

livebench.ai

Measures: 23 objective tasks across seven categories, refreshed every six months to resist contamination.

How to read it: it better reflects current reasoning than static tests such as MMLU because the questions are new. Scores from different releases are not directly comparable; compare models within one release.

Biases: task selection reflects the designers' preferences, and 23 tasks cannot represent every scenario.

Domain Benchmarks

General platforms assess overall quality; domain benchmarks answer whether a model can perform a specific job.

SWE-bench / SWE-bench Verified

swebench.com

Measures: real GitHub issues from 12 open-source Python projects including Django, Flask, and scikit-learn. A model reads the repository, fixes the bug, and must pass the associated tests.

Verified: the original contains ambiguous issues and weak tests. Verified is a manually reviewed 500-problem subset and the more frequently cited 2026 version.

How to read it: this is the toughest coding-agent benchmark. A high score shows repository understanding, bug localization, and correct patches. It measures only Python bug fixes—not new features, refactoring, or multilingual development.

Biases: Python only; accepts fixes that pass the original tests even when other valid fixes exist; vendors may train specifically on the coding style of these 12 repositories.

Aider Polyglot

aider.chat/docs/leaderboards

Measures: 133 multilingual exercises in Python, JavaScript, TypeScript, C#, Java, Go, and more. A model edits existing files until tests pass; API cost is recorded.

How to read it: closer than SWE-bench to everyday “change this code correctly” work. Cost is especially useful: GPT-5 (high) scored 88% for $29, while DeepSeek V3.2 Exp Chat scored 70% for $0.88.

Biases: Aider's prompt design affects results; the same model may behave differently in Cursor, Copilot, or Claude Code.

HumanEval

Measures: 164 Python tasks released by OpenAI in 2021. Given a signature and docstring, the model writes a function body that must pass tests.

How to read it: saturated by 2026, with leading models above 95%. It remains an introductory coding test but should not drive selection.

Biases: too easy, Python-only, and widely present in training corpora.

MMLU-Pro

Measures: a harder MMLU with about 12,000 questions across 14 subjects, ten answer choices rather than four, and more multi-step reasoning.

How to read it: random-guess accuracy falls from 25% to 10%, preserving separation among strong models. It still tests “what do you know?” through multiple choice, not “what can you do?”

Biases: multiple-choice format and dependence on training-data knowledge coverage.

GPQA Diamond

Measures: 198 graduate-level physics, chemistry, and biology questions written by specialists and validated as too hard for non-specialists even with Google.

How to read it: one of 2026's hardest science-reasoning tests. With only 198 items, small score changes may be random.

Biases: natural sciences only; no engineering, social science, or humanities; small sample and wide confidence intervals.

AIME (American Invitational Mathematics Examination)

Measures: 15 contest problems with integer answers from 0 to 999, requiring multi-step and creative reasoning.

How to read it: a clean math-reasoning test with little formatting bias; reports accuracy, such as AIME 2024 at 83.3%.

Biases: historical questions may enter training data. The 2024 set was newer but will also become contaminated.

Humanity's Last Exam (HLE)

Measures: extremely difficult questions across 100+ subjects, written by thousands of experts through Scale AI and CAIS and published in Nature (649, 1139–1146). Artificial Analysis evaluates 2,158 text-only items.

How to read it: progress is rapid. In April 2025, Gemini 3 Pro led at 38.3% and GPT-5 scored 25.3%. By August 2026, frontier models reached 55%+—Claude Fable 5 at 55.5% and Opus 5 at 54.9%—or 64.7% with tools. Nearly half remains unsolved.

Biases: results vary sharply by source. Artificial Analysis reports about 55% for Opus 5 on 2,158 text-only questions; BenchLM reports 64.7%. Different subsets, tools, judge models, or scoring can explain the gap. Always record the evaluator and conditions.

Multimodal Model Evaluation (VLM)

VLMs must understand images and text, adding another dimension to evaluation.

MMMU / MMMU-Pro

mmmu-benchmark.github.io

Measures: 11,500 university-level multimodal questions across 30 subjects, including charts, images, musical notation, and chemical structures. Pro expands choices from four to ten and removes questions answerable without the image.

How to read it: original MMMU saturated above 80% after 2025; Pro still separates models. Gemini 3.1 Pro scored 82%, GPT-5.4 81%, Qwen3.6 Plus 78.8%, and open Qwen3-VL-235B 69.3%, leaving closed models about 10pp ahead in broad academic reasoning.

MathVista

mathvista.github.io

Measures: 6,141 visual math problems involving chart arithmetic, geometry, and statistical tables.

How to read it: open models lead here: Qwen3-VL-235B scored 85.8%, above every closed model.

DocVQA / ChartQA / OCRBench

Measures: practical document understanding. DocVQA asks questions about scanned invoices, contracts, and tables; ChartQA tests charts; OCRBench v2 tests bilingual recognition.

How to read it: these matter more than MMLU for enterprise document pipelines. Qwen2.5-VL-72B scored 96.4% on DocVQA, above GPT-5.4 at 95%.

Image Generation Evaluation

Images have no single objective answer, so evaluation relies more on human preference and multidimensional automated scoring.

LMArena Text-to-Image Arena

arena.ai/leaderboard/text-to-image

Measures: pairwise blind ELO voting on images generated from the same prompt, across 76+ models.

August 2026 top five: GPT Image 2 (1381), MAI-Image-2.6 (1336), Grok Imagine 2.0 (1316), Reve 2.1 (1302), Meta Muse Image (1282).

Artificial Analysis Image Arena

artificialanalysis.ai/image/leaderboard/text-to-image

Measures: independent image ELO plus API price per image, combining quality and cost. GPT Image 2 high costs $211/thousand images versus $48 for MAI-Image-2.5.

Evalytic

evalytic.ai/leaderboard

Measures: 33 models, 100 prompts, and six AI judges—CLIP Score, PickScore, HPSv2, ImageReward, VQAScore, and a VLM judge—with subdimensions for visual quality, prompt adherence, and text rendering.

How to read it: subdimensions help when a specific ability matters, such as rendering text, more than Arena's single ELO.

Biases: AI preferences differ from human preferences. CLIP Score, NIMA, and sharpness capture technical quality rather than aesthetics.

Traditional Metrics: FID / CLIP Score

FID measures distribution similarity between generated and real images; CLIP Score measures image-text alignment. They dominated before 2024, but by 2026 their correlation with human preference was considered too weak for primary use. Treat them as supporting metrics.

Video Generation Evaluation

Video was the most volatile 2026 category, with rankings changing every few weeks.

Artificial Analysis Video Arena

artificialanalysis.ai/video/leaderboard/text-to-video

Measures: blind-vote ELO for text-to-video and image-to-video, each split into with-audio and without-audio boards, covering 30+ models with $/minute prices.

August 2026 top five, T2V with audio: Wan 3.0 (1244), Gemini Omni Flash (1238), MiniMax H3 (1228), Seedance 2.0 (1221), Wan 2.7 (1156).

Biases: rankings fluctuate more than text or image rankings. Kling 3.0 1080p Pro moved by 137 ELO and three positions across measurements. Always date video rankings.

LMArena Image-to-Video Arena

arena.ai/leaderboard/image-to-video

Measures: ELO across 45 I2V models. MiniMax H3 led at 1489 in August 2026.

VBench / VBench-2.0

vchitect.github.io/VBench-project

Measures: VBench decomposes quality into 16 dimensions such as subject consistency, motion smoothness, temporal flicker, and aesthetics. VBench-2.0 added 18 physical-realism dimensions in March 2026, including fluid flow and collision inertia.

How to read it: Arena measures subjective preference; VBench measures technical quality, and they can disagree. Open Wan 2.2 scored 84.7% on VBench, while even leading models scored only about 50% on VBench-2.0's physical-motion dimensions.

Speech and Music Evaluation

Artificial Analysis Speech Arena

artificialanalysis.ai/text-to-speech/arena

Measures: blind-listening ELO between two TTS outputs of the same text. Qwen-Audio-3.0-TTS ranked first in July 2026.

TTS-Bench

github.com/5uck1ess/tts-bench

Measures: 65 TTS models across speed (TTFA, RTFx), blind A/B listening, and objective UTMOS naturalness, WER clarity, and SIM voice similarity, on RTX 5090, M4, and RTX 3090 hardware.

How to read it: the most practical local-deployment benchmark because it reports speed and naturalness by hardware. Its public TTS Voting Arena collects blind votes.

MINT-Bench

arxiv.org/abs/2604.17958

Measures: multilingual instruction-following TTS across ten languages and three layers: content consistency, instruction following, and perceptual quality. Frontier commercial systems lead in English, while Qwen3-TTS beats Gemini in Chinese.

Speech Recognition (ASR) Metrics

WER is the standard ASR metric. The OpenASR Leaderboard evaluates LibriSpeech, Common Voice, and other datasets. Whisper Large V3 remains the most widely used baseline.

Music Generation Evaluation

Evaluation remains early:

  • FAD: audio equivalent of FID, comparing generated and real music distributions.
  • MusicCaps: Google's music-description dataset for text-to-music alignment.
  • SongBench (Tencent, 2026): the largest song benchmark, with 11,717 expert annotations across vocals, instruments, melody, structure, arrangement, mixing, and musicality.

Music is highly subjective; musicians and general listeners can disagree sharply on the same generated song.

Embedding and Rerank Evaluation

MTEB Leaderboard

huggingface.co/spaces/mteb/leaderboard

Measures: embedding quality across retrieval, classification, clustering, and semantic similarity; the standard selection leaderboard.

How to read it: separate multilingual from English-only results. BGE-M3 leads multilingual work; Qwen3-Embedding and Voyage are strong on code.

BEIR

github.com/beir-cellar/beir

Measures: cross-domain retrieval, with reranker accuracy measured by nDCG@10. Main competitors include Jina Reranker v3.5 at 63.2, Qwen3-Reranker-8B, and BGE-Reranker.

How to read it: the 2026 RAG standard is BGE-M3 or Qwen3-Embedding for recall plus a reranker. BEIR nDCG@10 is the primary reranker metric.

Market Adoption Indicators

Benchmarks measure capability; adoption measures how many people use a model. They need not agree.

OpenRouter Rankings

openrouter.ai/rankings

Measures: actual prompt plus completion token volume across hundreds of models through one aggregation API. Data is public under CC BY 4.0 and available through a Data API.

August 2026 top ten, through Aug. 23:

RankModelToken volume
1DeepSeek V4 Flash 073111.6T
2Ox Alpha (stealth)11.6T (new)
3MiMo-V2.5 (xiaomi)9.94T
4Hy3 (tencent)8.21T
5DeepSeek V4 Flash 04235.46T
6GPT-5.6 Luna4.91T
7Nemotron 3 Ultra (free)4.75T
8GLM 5.23.42T
9DeepSeek V4 Pro 04231.85T
10Gemini 3.7 Flash1.8T

How to read it: this is the closest indicator of market adoption, not quality—OpenRouter explicitly says so. Strong benchmarks with little usage may indicate high price or an unstable API; high usage without top scores often indicates value or a mature ecosystem.

Biases: users skew technical; enterprise and direct Anthropic/OpenAI API usage is missing; tokenizers differ, so token counts are not perfectly comparable; free tiers such as Nemotron 3 Ultra may inflate use.

Hugging Face Downloads / Likes

huggingface.co/models

Measures: downloads and likes for open models, filterable by pipeline tag, seven-day trend, and lifetime totals.

How to read it: downloads include CI/CD pulls and can exceed actual users; likes measure attention rather than durable adoption; Trending Score is a seven-day weighted signal of what is hot now. Quantizations often out-download originals because more people can run them, not because they are better.

Biases: open models only; a download does not prove deployment.

Ollama Library

ollama.com/library

Measures: pulls and ranking in the most popular local LLM runner.

How to read it: the best indicator of what people run on personal computers, usually models that balance quality with consumer-hardware performance.

Biases: personal and small-team use only; enterprise vLLM, TGI, and SGLang deployments are absent.

Official Sources

Vendor Model Cards / Blogs

Launch posts contain benchmarks, architecture, and prices. They are primary sources but require the most caution.

How to read them: mark self-tests with a warning because prompt templates, temperature, few-shot setup, and post-processing change results. Compare only independent runs using one protocol. Missing benchmarks may be selectively omitted. Distinguish “reaches Y%” from “beats every model.” Confirm pricing on the pricing page, including standard, batch, and cached rates, rather than launch promotions.

VendorAnnouncementsPricing
Anthropicanthropic.com/newsanthropic.com/pricing
OpenAIopenai.com/blogopenai.com/api/pricing
Googleblog.google/technology/aiai.google.dev/pricing
Metaai.meta.com/blogOpen source; no API price
xAIx.ai/blogdocs.x.ai
Mistralmistral.ai/newsmistral.ai/products
DeepSeekapi-docs.deepseek.com/newsapi-docs.deepseek.com
Qwenqwenlm.github.io/blogOpen source + Tongyi API
Coherecohere.com/blogcohere.com/pricing

Common Traps

Benchmark Gaming

Vendors optimize for watched benchmarks by including identical or similar questions in training, special-casing known formats at inference, or selectively reporting favorable versions. Defense: use independent reproduction and contamination-resistant tests such as LiveBench and HLE.

Leaderboard Saturation

HumanEval and original MMLU are saturated above 95%. A 0.5-point gap is not practically meaningful even when marketing says “beats competitors.” Defense: inspect confidence intervals; overlapping intervals imply no demonstrated difference.

Self-Reported Numbers

Vendor and independent scores differ because of prompts, temperatures, few-shot settings, and post-processing. Defense: compare one benchmark only within one evaluator. Anthropic's MMLU-Pro run is not directly comparable with OpenAI's.

Cherry-Picking

“Beats GPT-4o” may compare against a six-month-old version rather than the latest. Defense: confirm version and date; GPT-4o could mean 2024-08-06, 2025-02-15, or chatgpt-4o-latest, with substantial performance differences.

Stale Data

A March 2026 comparison may be three model generations behind by August. Defense: check publication date and model version; prefer continuously updated sources such as Artificial Analysis, LMArena, LiveBench, and Aider.

Selection Framework: What to Read by Scenario

Different work requires different metrics.

Agents / Tool Calling

PrioritizeWhat it tells you
SWE-bench VerifiedMulti-step task completion
Aider PolyglotCode editing success + cost
Artificial Analysis Coding Agent IndexEnd-to-end agent performance
tau-benchMulti-turn tool success (vendor self-tests)
PrioritizeSource
MTEBEmbedding quality
BEIR nDCG@10Reranker accuracy
Context windowVendor model card
RULER / Needle in a HaystackLong-context retrieval

Cost / High Throughput

PrioritizeSource
Artificial Analysis price + speedStandardized $/1M and tokens/sec
Aider costActual cost for identical tasks
OpenRouter RankingsHigh-use models often offer good value

General Chat / Customer Service

PrioritizeSource
LMArena ELOReal user preference
MMLU-ProKnowledge breadth
Multilingual benchmarks such as AyaNon-English service

Reasoning / Math / Science

PrioritizeSource
AIMEMath reasoning
GPQA DiamondScience reasoning
HLEExtreme cross-domain reasoning
LiveBench ReasoningContamination-resistant reasoning

Image Generation

PrioritizeSource
LMArena / AA Image ArenaHuman-preference ELO
Evalytic TR / PAText rendering and prompt adherence
$/1,000 imagesArtificial Analysis pricing

Video Generation

PrioritizeSource
AA Video Arena, datedVolatile human-preference ELO
VBench / VBench-2.016+18 technical dimensions
$/minuteProvider pricing

Text-to-Speech (TTS)

PrioritizeSource
AA Speech ArenaHuman-preference ELO
TTS-BenchLocal speed, listening, and objective scores
SEED-TTS-EvalVoice-cloning quality

Multimodal Understanding (VLM)

PrioritizeSource
MMMU-ProCross-subject visual reasoning
DocVQA / ChartQAPractical document understanding
MathVistaVisual math reasoning

Overall

No single benchmark tells you whether a model is good. Instead:

  1. Define the scenario and select two or three relevant metrics.
  2. Use independent evaluations, not vendor self-tests.
  3. Cross-check quality benchmarks, adoption, and your own tests.
  4. Always attach a date and model version to every number.

The greatest risk is not choosing the wrong model. It is choosing with the wrong numbers.


References