OpenGPTAI decision intelligence
Search⌘K
OFFICIAL MODEL REFERENCE

AI model directory

Browse exact model IDs listed in provider documentation. Only versions reliably linked to comparable public data enter the rankings.

Model series
13
Official versions
79
Public ranking available
42
Awaiting comparable data
37
OFFICIAL MODEL REFERENCE

Official model IDs

Every tracked version remains visible. Ranked versions identify their data sources; other versions explain why no ranking is shown.

13 series · 79 versionsSubmit model
Deployment
Ranking data
13 / 13 series · 79 versions

Showing 2 of 13 model series

01 · 7 versions

GPT

A broad proprietary model family with a mature API and tool ecosystem.

Everyday workSolve problems or codeUse current information or tools
Official source

Compare versions

3 of 7 versions shown · Reviewed Jul 2026

Start with this series →
GPT · Compare versions
Select for comparisonModel / versionBest fitPublic ranking dataOfficial access
gpt-5.6-solHighest capability
RankedComparable public data is linked to this exact version.
User preference 18 ↗Overall capability 5 ↗Objective tasks 2 ↗
Ranked with a special source configuration that site comparison cannot reproduce yet.
Model detailsOfficial access ↗
gpt-5.6-terraQuality and cost balance
Not yet rankedNo comparable public result is available for this exact version in the current edition.
Model detailsOfficial access ↗
gpt-5.6-lunaSpeed and volume
Not yet rankedNo comparable public result is available for this exact version in the current edition.
Model detailsOfficial access ↗

Strengths

Covers general, coding, vision, and tool-based workflows.

Limitations

Closed weights and provider-controlled deployment.

02 · 6 versions

Claude

A proprietary family positioned for analysis, writing, coding, and long-context work.

Read, analyze, or writeRead long documentsSolve problems or code
Official source

Compare versions

3 of 6 versions shown · Reviewed Jul 2026

Start with this series →
Claude · Compare versions
Select for comparisonModel / versionBest fitPublic ranking dataOfficial access
claude-fable-5Highest capability
RankedComparable public data is linked to this exact version.
User preference 1 ↗Overall capability 3 ↗Objective tasks 1 ↗
Model detailsOfficial access ↗
claude-sonnet-5Quality and cost balance
RankedComparable public data is linked to this exact version.
User preference 44 ↗
Ranked with a special source configuration that site comparison cannot reproduce yet.
Model detailsOfficial access ↗
claude-opus-4-8Coding and agents
RankedComparable public data is linked to this exact version.
User preference 26 ↗
Model detailsOfficial access ↗

Strengths

Well suited to document analysis and sustained written work.

Limitations

Closed weights and API-dependent access.

Showing 2 of 13 model series
PUBLIC SOURCE INDEX

Complete source-listed model index

All 418 published records from 3 public sources, grouped into 414 unique source labels. Inclusion is not a recommendation; identity mappings are shown separately.

418 / 414
Published records
418
Unique source labels
414
Mapped to official IDs
54
Source-only labels
360
Public source
Identity status

Showing 12 of 414 source labels

Artificial AnalysisSource-only label
Source model ID

Claude Opus 5

Anthropic

Overall capability · #1 · 63Overall capability · #2 · 63Overall capability · #4 · 61
Artificial AnalysisSource-only label
Source model ID

Kimi K3

Kimi

Overall capability · #7 · 60
Artificial AnalysisSource-only label
Source model ID

Grok 4.6

SpaceXAI

Overall capability · #6 · 61
LiveBenchSource-only label
Source model ID

Smaug-Agentic

Abacus.AI

Objective tasks · #6 · 79.7Open weights · #1 · 79.7
LiveBenchSource-only label
Source model ID

Qwen 3.8 Max

Alibaba

Objective tasks · #7 · 79.5Open weights · #2 · 79.5
LiveBenchSource-only label
Source model ID

Claude Opus 5

Anthropic

Objective tasks · #4 · 80.5
LiveBenchSource-only label
Source model ID

DeepSeek V4 Flash 0731

DeepSeek

Cost per successful task · #5 · $0.0596
Showing 12 of 414 source labels