OpenGPTEvidence-based model decisions

Search OpenGPT

Models, agents, rankings, and tools

Go toAI model rankingsPublic-source model rankingsGo toAI Agent directoryProducts and comparable public evaluationsGo toChoose a modelStart from your task and build a shortlistGo toTask rankingsBrowse evidence by real-world taskGo toCompare modelsCompare exact versions and key differencesGo toReal-task evaluationBlind-review answers on the same tasksGo toRanking changesSee version and ranking historyGo toModel directoryBrowse official model IDs
Start evaluation

AI model rankings

Submit model

Compare exact model configurations across independent user-preference, intelligence, objective-task, cost, and open-weight-model rankings—then test two on your own work.

Start with your taskCompare two leading modelsLatest model changes
Published evaluation data3 external public sourcesEdition data cutoff 2026-08

OpenGPT organizes published third-party evaluation data; it did not rerun these tests. Use the rankings to shortlist models, then validate them on your own work.

Method version
OpenGPT research synthesis 1.1
Ranking method

Choose a ranking metric

Start with the decision you need to make. Each scenario opens the most relevant published ranking signal. The original source and scale remain visible.

Choose what to show

Show every publisher record with its source model ID or source name, including aliases and run configurations. Source ranks are preserved; ties can still cause rank numbers to skip.

User preference score. LMArena’s complete latest style-controlled overall ranking. Every publisher record keeps its source model ID; a provider-documented model ID appears separately when available.

Compared with previous edition Jul 21, 2026. Same source and method. Rank movement is relative; it does not by itself prove a change in model quality.

Showing 10 of 385. User preference score.

  1. Rank 1

    claude-fable-5

    Anthropic
    Model ID
    claude-fable-5
    Votes cast
    19,390
    License type
    Closed / proprietaryProprietary
    User preference score1507Based on anonymous head-to-head comparisons
    Likely score range1,501.7–1,512.9
    Model detailsReview match evidenceView source
  2. Rank 2

    claude-opus-4-6-thinking

    Anthropic
    Model ID
    claude-opus-4-6
    Votes cast
    69,336
    License type
    Closed / proprietaryProprietary
    User preference score1505Based on anonymous head-to-head comparisons
    Likely score range1,501.0–1,508.2
    Model detailsReview match evidenceView source
  3. Rank 3

    claude-opus-4-7-thinking

    Anthropic
    Model ID
    claude-opus-4-7
    Votes cast
    57,018
    License type
    Closed / proprietaryProprietary
    User preference score1502Based on anonymous head-to-head comparisons
    Likely score range1,497.7–1,505.7
    Model detailsReview match evidenceView source
  4. Rank 4

    muse-spark-1.2 (xHigh)

    Meta
    Source model ID
    muse-spark-1.2 (xHigh)
    Votes cast
    2,057
    License type
    Closed / proprietaryProprietary
    User preference score1498Based on anonymous head-to-head comparisons
    Likely score range1,484.7–1,510.8
    Model detailsView source
  5. Rank 5

    claude-opus-4-6

    Anthropic
    Model ID
    claude-opus-4-6
    Votes cast
    73,158
    License type
    Closed / proprietaryProprietary
    User preference score1497Based on anonymous head-to-head comparisons
    Likely score range1,494.0–1,500.9
    Model detailsReview match evidenceView source
  6. Rank 6

    qwen3.8-max

    Alibaba
    Source model ID
    qwen3.8-max
    Votes cast
    4,662
    License type
    Closed / proprietaryProprietary
    User preference score1497Based on anonymous head-to-head comparisons
    Likely score range1,488.3–1,506.0
    Model detailsView source
  7. Rank 7

    claude-opus-4-7

    Anthropic
    Model ID
    claude-opus-4-7
    Votes cast
    58,135
    License type
    Closed / proprietaryProprietary
    User preference score1493Based on anonymous head-to-head comparisons
    Likely score range1,488.9–1,496.8
    Model detailsReview match evidenceView source
  8. Rank 8

    claude-opus-5-high

    Anthropic
    Source model ID
    claude-opus-5-high
    Votes cast
    14,902
    License type
    Closed / proprietaryProprietary
    User preference score1493Based on anonymous head-to-head comparisons
    Likely score range1,486.9–1,498.6
    Model detailsView source
  9. Rank 9

    claude-opus-5-max

    Anthropic
    Source model ID
    claude-opus-5-max
    Votes cast
    7,137
    License type
    Closed / proprietaryProprietary
    User preference score1488Based on anonymous head-to-head comparisons
    Likely score range1,480.7–1,496.2
    Model detailsView source
  10. Rank 10

    muse-spark

    Meta
    Model ID
    muse-spark
    Votes cast
    13,476
    License type
    Closed / proprietaryProprietary
    User preference score1488Based on anonymous head-to-head comparisons
    Likely score range1,482.1–1,493.9
    Model detailsReview match evidenceView source
Showing 10 of 385
Independent sourceLMArena385 source-listed model entries · Data retrieved: Aug 6, 2026
Leaderboard publication date
Aug 6, 2026
Ranking scope
Text responses · Style-controlled · Overall ranking
Data retrieved
Aug 6, 2026
Original ranking ↗Scoring method ↗
LMArena

Monitoring statusLive status unavailableValidated automatic publication

Source data date
OpenGPT retrieved
Ranking last changed

LMArena: validated automatic publication · Artificial Analysis and LiveBench: manual review

Ranking methodEach view preserves the source’s metric, release, and methodology. Follow the links to audit the underlying work.

Ranking basis

Full published ranking and official model IDs

The main ranking preserves every publisher record and shows its source model ID or source name. Every row can be compared: exact, configuration-safe mappings use official versions; all other rows stay within one source ranking. Missing data never becomes a zero score.
79 official model IDs
79
44 with public ranking data
44
35 awaiting comparable data
35
385 source records retained for audit
385
Model directory

Ranking basis

How to read this research edition

  • Scores stay on their original publisher scales. They are not combined into a composite score.

  • Missing data does not mean zero. It means the publisher did not report that metric for this snapshot.

  • When a source reports model settings, the position applies to that exact setup and source release—not every task, region, or deployment.

  • No provider can pay for placement. OpenGPT links directly to the evidence used.

Ranking basis

Data and sources

Each view preserves the source’s metric, release, and methodology. Follow the links to audit the underlying work.
AR

LMArena

The complete latest style-controlled overall ranking, based on users comparing two model answers side by side.

Leaderboard publication date
Aug 6, 2026
Data retrieved
Aug 6, 2026
License
CC BY 4.0

Original ranking ↗Scoring method ↗

AA

Artificial Analysis

Nine evaluations across agentic work, coding, general knowledge, and scientific reasoning.

Source snapshot
Intelligence Index v4.1.1 · continuously updated
Data retrieved
Aug 6, 2026

Original ranking ↗Scoring method ↗

LB

LiveBench

Twenty-three objective tasks across seven categories; questions refresh every six months.

Source snapshot
LiveBench-2026-06-25
Data retrieved
Aug 8, 2026

Original ranking ↗Scoring method ↗

Method version

A predictable publishing rhythm

OpenGPT checks model sources every six hours. Validated LMArena updates can publish automatically; Artificial Analysis and LiveBench require manual review. Failed validation keeps the previous ranking active.
  1. 01

    6-hour source checks

    LMArena updates publish automatically only after schema, completeness, date, and anomaly checks. Artificial Analysis and LiveBench stay under manual review.

  2. 02

    Weekly change summary

    Summarize meaningful ranking, model, and source changes in one reviewable update.

  3. 03

    Monthly history snapshot

    Freeze a dated snapshot with exact configurations, source releases, and corrections.