Live model ranking radar

ModelRank

A practical ranking page for image, video, coding, audio, OCR/document, and vision models. It prioritizes public APIs and raw CSVs, then uses HTML-only leaderboards as cross-checks.

Source LMArena Text-to-Image Loading live ranking data...

Original source

Ranking source

Rank	Model	Organization	Score	Access	Evidence
Loading live ranking data...

Data sources

Every source below is tagged by sync method and confidence so future Codex updates can decide what to automate or keep as a reference.

API sync High confidence

LMArena Text-to-Image

Best automatic source for mixed open/closed image-generation rankings.

Metric: Arena rating, higher is better
Cadence: Sync every 30 minutes; upstream latest split updates when LMArena republishes.

Open and closed models mixed Original source

API sync High confidence

LMArena Image Edit

Complements text-to-image with editing models such as GPT Image, Gemini image, Seedream, and others.

Metric: Arena rating, higher is better
Cadence: Sync every 30 minutes.

Open and closed models mixed Original source

API sync High confidence

LMArena Text-to-Video

Good primary source for text-to-video, including ByteDance, Kuaishou, xAI, Google, and other providers.

Metric: Arena rating, higher is better
Cadence: Sync every 30 minutes.

Open and closed models mixed Original source

API sync High confidence

LMArena Image-to-Video

Secondary video source for image-to-video workflows.

Metric: Arena rating, higher is better
Cadence: Sync every 30 minutes.

Open and closed models mixed Original source

API sync High confidence

SWE-bench Verified

Best structured source for agentic coding on real GitHub issues.

Metric: Resolved percentage, higher is better
Cadence: Sync hourly; source is a Hugging Face benchmark leaderboard API.

Open and closed models mixed Original source

API sync High confidence

LMArena WebDev

Complements SWE-bench with web-development preference rankings.

Metric: Arena rating, higher is better
Cadence: Sync every 30 minutes.

Open and closed models mixed Original source

API sync High confidence

Open ASR Leaderboard

Strong source for speech recognition. Audio generation leaderboards still need a secondary curated source.

Metric: Average WER, lower is better
Cadence: Sync daily or hourly; benchmark API is structured and reproducible.

Open and closed models mixed Original source

CSV sync Medium confidence

OCRBench v2 English

Useful for OCR and text-centric visual understanding; treat as medium-confidence until mirrored into D1 snapshots.

Metric: Average score, higher is better
Cadence: Sync daily from raw CSV; upstream cadence is less formal than HF benchmark API.

Open and closed models mixed Original source

CSV sync Medium confidence

OCRBench v2 Chinese

Chinese OCR and document-understanding companion table.

Metric: Average score, higher is better
Cadence: Sync daily from raw CSV.

Open and closed models mixed Original source

API sync High confidence

LMArena Document

Good live source for document-style multimodal tasks.

Metric: Arena rating, higher is better
Cadence: Sync every 30 minutes.

Open and closed models mixed Original source

API sync High confidence

LMArena Vision

Broad visual reasoning and multimodal model comparison.

Metric: Arena rating, higher is better
Cadence: Sync every 30 minutes.

Open and closed models mixed Original source

HTML watch Watch source

Artificial Analysis Video

High-quality public leaderboard, but parsing HTML is more fragile than using benchmark APIs.

Metric: Video Arena ELO, higher is better
Cadence: Use as a cross-check unless a stable public API is available.

Open and closed models mixed Original source

HTML watch Watch source

Aider Polyglot

Good benchmark for editing across C++, Go, Java, JavaScript, Python, and Rust, but no stable public JSON found.

Metric: Pass rate, higher is better
Cadence: Manual or HTML-parse fallback; useful as a code-editing cross-check.

Open and closed models mixed Original source