AI BENCHMARKS

Best at what?

Reasoning, writing, building websites and generating media are different jobs. Here’s how the leaders compare in each.

Results from Artificial Analysis and Arena. Artificial Analysis refreshes daily only while its methodology matches the reviewed v5.0.1 reference. Arena scores are reviewed in a browser. Check source dates, model settings and uncertainty when comparing scores.

Artificial Analysis

Reasoning & knowledge Updated

Checked 2026-11-18

A composite of knowledge work, coding, science and general evaluations. Higher scores are better; these are index points, not percent correct.

Intelligence Index v5.0.1

  1. Claude Fable 5.2 Max, Default FallbackIntelligence Index
    65
  2. GPT-6.2 Astra MaxIntelligence Index
    62
  3. Gemini 4 Carbon HighIntelligence Index
    60
  4. Claude Opus 5.5 Max, Default FallbackIntelligence Index
    59
  5. GPT-6.2 Sol MaxIntelligence Index
    57

Effort settings and fallback runs affect results. An aggregate score can hide weaknesses on your particular task. Version 5.0 changed the evaluation suite and weighting (5.0.1 only regraded one test); compare only scores from the same major version.

Full results at Artificial Analysis ↗

Arena

Writing & conversation Updated

Checked 2026-11-10 · Browser-reviewed snapshot · Source results older than 7 days

People compare anonymous model responses and choose the answer they prefer.

Overall · Style Control · All licenses

Source date shown by Arena: 2026-11-09. Checked dates record our review, not a new source release.

  1. Gemini 4 Carbon (High)Preliminary · 4,282 votes
    1526±10
  2. Gemini 4 Argon (High)19,787 votes
    1522±5
  3. Muse Wonder (Max)12,032 votes
    1511±6
  4. Claude Opus 5.5 (High)15,249 votes
    1506±6
  5. Claude Fable 5 (High)45,524 votes
    1504±4

Preference is not a factual-accuracy score. A listing does not guarantee access: some entries are previews, and others need a paid plan. Uncertainty ranges overlap for many entries; their order is not proof of a meaningful difference.

Full results at Arena ↗

Arena

Web development Updated

Checked 2026-11-10 · Browser-reviewed snapshot · Source results older than 7 days

Head-to-head preferences for generated web applications.

WebDev Overall · All licenses

Source date shown by Arena: 2026-11-10. Checked dates record our review, not a new source release.

  1. Claude Opus 5.5 (Max)4,166 votes
    1813+12/-12
  2. Claude Sonnet 5.5 (xHigh)5,778 votes
    1772+10/-10
  3. Gemini 4 Carbon (High)Preliminary · 2,124 votes
    1768+15/-15
  4. GPT-6 Astra (Max)8,390 votes
    1765+9/-9
  5. GPT-6.1 Sol (Max)5,512 votes
    1743+10/-10

This measures web-building output, not every kind of coding, security or maintenance work. Uncertainty ranges overlap for many entries; their order is not proof of a meaningful difference.

Full results at Arena ↗

Arena

Image generation Updated

Checked 2026-11-10 · Browser-reviewed snapshot · Source results older than 7 days

Human preferences between images generated from prompts.

Text-to-image Overall · All licenses

Source date shown by Arena: 2026-11-07. Checked dates record our review, not a new source release.

  1. GPT-Image-2.5 SunburstPreliminary · 57,928 votes
    1429±5
  2. GPT-Image-2.6 Sunburst (Medium)Preliminary · 2,718 votes
    1406±13
  3. GPT-Image-2.5 FlarePreliminary · 57,020 votes
    1400±5
  4. GPT-Image-2.6 Flare (Medium)Preliminary · 2,551 votes
    1383±13
  5. GPT-Image-2 (Medium)106,021 votes
    1382±4

A high rating does not guarantee accurate text, product details or permission to use an image commercially. Uncertainty ranges overlap for many entries; their order is not proof of a meaningful difference.

Full results at Arena ↗

Arena

Video generation Updated

Checked 2026-11-10 · Browser-reviewed snapshot · Source results older than 7 days

Human preferences between generated video clips.

Text-to-video Overall · All licenses

Source date shown by Arena: 2026-11-07. Checked dates record our review, not a new source release.

  1. Gemini Omni 1.2 FlashPreliminary · 1,714 votes
    1540±16
  2. Gemini Omni 1.1 Flash1,792 votes
    1513±15
  3. Gemini Omni Flash46,707 votes
    1509±8
  4. FLUX 3 VideoPreliminary · 1,314 votes
    1491±17
  5. Grok Imagine Video 1.5Preliminary · 1,249 votes
    1488±19

Compare duration, resolution and audio separately. Early results can move substantially as votes accumulate. Uncertainty ranges overlap for many entries; their order is not proof of a meaningful difference.

Full results at Arena ↗

Read the result, not just the rank.

Keep settings attached

“High,” “max,” tool access and fallback runs can change both results and cost. They are part of the tested configuration.

Watch uncertainty

A small rating gap with overlapping ranges is weak evidence of a clear winner. Preliminary entries may have fewer votes.

Compare within a board

An Arena rating of 1,500 cannot be compared with an Intelligence Index of 50. The scales and underlying tasks differ.

Try your own workload

Use known answers, count retries and compare time to a usable result. Long context is capacity, not a guarantee of accurate recall.

Compare access and API costs → · A practical benchmark guide →