LITHOVEX AI · Model discovery
Explore AI models with clear catalog comparisons
Review model profiles and provider listings, compare catalog records, and use the available details to decide which models to test for your work. The current model catalog includes 249 model guides. Review listed specifications, compare shared benchmark entries, and open a model profile to check its sources and details.
Featured model guides
Catalog coverage: 249 model records across 83 listed providers; 238 records include benchmark entries. Coverage reflects fields present in the catalog and is not a ranking of model quality.
- DeepSeek-R1 · DeepSeek
- Kimi-K2 Instruct · Moonshot AI
- Qwen3 235B-A22B (MoE) · Qwen
- DeepSeek-R1 Distill Qwen 32B · DeepSeek
- DeepSeek-V3 · DeepSeek
- DeepSeek-V3 (Mar 24) · DeepSeek
- DeepSeek-V3.1 · DeepSeek
- Google Gemini 3.1 Pro · Google
- OpenAI GPT-5.4 Pro · OpenAI
- Qwen3 32B · Qwen
- Anthropic Claude Opus 4.6 (Pro) · Anthropic
- Anthropic Claude Opus 4.7 · Anthropic
Browse the AI model directory or model comparisons.
Examples from listed benchmarks
- DeepSeek-R1 — MMLU: 90.8%
- DeepSeek-R1 — HumanEval: 92.0%
- Kimi-K2 Instruct — SWE-bench Verified: 65.8%
- Kimi-K2 Instruct — LiveCodeBench: 75.2%
- Qwen3 235B-A22B (MoE) — MMLU: 93.5%
- Qwen3 235B-A22B (MoE) — HumanEval: 96.2%
- DeepSeek-R1 Distill Qwen 32B — AIME 2024: 72.6%
- DeepSeek-R1 Distill Qwen 32B — MATH-500: 94.3%
Benchmark names, versions, and evaluation setups can differ. These catalog entries are reference data, not a controlled head-to-head evaluation.
Browse by provider
- AI Singapore
- akhaliq
- ali-vilab
- AllenAI
- Amazon
- Anthropic
- Arcee AI
- Baidu
- bigcode
- Blackbox AI
- briaai
- caidas
- cardiffnlp
- cerspense
- codellama
- Cohere
- Community
- coqui
- DeepSeek
- deepseek-ai
- defog
- depth-anything
- Dicta
- distil-whisper
- espnet
- gorilla-llm
- gradientai
- Helsinki NLP
- HuggingFaceH4
- HuggingFaceM4
- hustvl
- impira
- infly
- KAIST AI
- LITHOVEX
- llava-hf
- MCG-NJU
- meetkai
- Meta Llama
- meta-llama
- microsoft
- Microsoft / OpenAI
- MiniMax
- Mistral AI
- mistralai
- mit
- Moonshot AI
- Morph
- myshell-ai
- naver-clova-ix
- nielsr
- nlpconnect
- nlptown
- Nous Research
- NVIDIA
- openai
- OpenAssistant
- openchat
- PekingU
- philschmid
- pszemraj
- pyannote
- Qwen
- Salesforce
- Sao10K
- SebastianBodza
- stabilityai
- suno
- tiiuae
- togethercomputer
- Trinity
- vennify
- vikp
- WizardLM
- xAI
- Zhipu AI
Browse all model categories
Selected model comparisons
Frequently asked questions
Are the listed benchmark results a direct head-to-head test?
No. Catalog benchmark records may use different versions and evaluation setups; verify results with the source provider before relying on them.
How should I compare two models?
Review the recorded specifications and test both models on the same representative task. Catalog entries are reference information, not a universal ranking.
Read About LITHOVEX AI and Contact, or review the Privacy Policy and Terms of Service. Visit the LITHOVEX Agency or read the model data methodology.