A curated collection of AI model cards, specs, comparison guides, and practical resources

| Model |
Org |
Release |
Params |
Highlights |
Card |
| GPT-5.2 |
OpenAI |
Jan 2026 |
~1-2T |
100% AIME, 400K context |
📄 |
| Gemini 3 Pro |
Google |
Jan 2026 |
MoE |
1M context, multimodal leader |
📄 |
| Llama 4 |
Meta |
Feb 2026 |
400B MoE |
Open weights, 52B active |
📄 |
| Doubao-Seed-2.0 |
ByteDance |
Feb 2026 |
Unknown |
Competes with GPT-5.2, Gemini 3 Pro |
📄 |
| Kimi K2 |
Moonshot AI |
Jul 2025 |
1T (32B active) |
MoE, Muon optimizer, agentic focus |
📄 |
| Claude Opus 4 |
Anthropic |
2025 |
Unknown |
200K context, extended thinking |
📄 |
| DeepSeek R1 |
DeepSeek |
Jan 2025 |
671B MoE |
Open source, matches o1, MIT license |
📄 |
| Grok 4 |
xAI |
Late 2025 |
Unknown |
Real-time X data, Aurora image gen |
📄 |
| Qwen 3 |
Alibaba |
2025 |
Up to 235B |
Apache 2.0, excellent value |
📄 |
| Model |
Org |
Release |
Resolution |
Highlights |
Card |
| Seedance 2.0 |
ByteDance |
Feb 2026 |
2K |
4-modal input, native audio sync |
📄 |
| Sora 2 |
OpenAI |
— |
1080p |
Coming soon |
|
| Kling 3.0 |
Kuaishou |
— |
1080p |
Coming soon |
|
| Veo 3.1 |
Google |
— |
1080p |
Coming soon |
|
| Model |
Org |
Release |
Highlights |
Card |
| FLUX 1.1 Pro |
Black Forest Labs |
Oct 2024 |
Best quality, great text rendering, 6x faster |
📄 |
| DALL-E 4 |
OpenAI |
— |
Coming soon |
|
| Imagen 3 |
Google |
— |
Coming soon |
|
| Midjourney v6 |
Midjourney |
— |
Coming soon |
|
| Model |
Type |
Card |
| Whisper v4 |
Speech Recognition |
📄 |
| text-embedding-3 |
Text Embeddings |
📄 |
| Use Case |
Top Pick |
Runner-up |
| Complex reasoning |
Claude Opus 4 |
DeepSeek R1 |
| Coding assistant |
Claude Opus 4 |
DeepSeek R1 |
| Budget-friendly |
DeepSeek V3 |
Qwen 3 |
| Self-hosting |
DeepSeek R1 (MIT) |
Qwen 3 (Apache) |
| Image generation |
FLUX 1.1 Pro |
DALL-E 3 |
| Video generation |
Seedance 2.0 |
Sora 2 |
| Agentic/tools |
Kimi K2 |
Claude Opus 4 |
| Budget |
Model |
Cost (per 1M tokens) |
| 💰 Free |
Qwen 3 (self-host) |
$0* |
| 💰 Low |
DeepSeek V3 |
$0.27 input |
| 💰💰 Medium |
Claude Sonnet 4 |
$3 input |
| 💰💰💰 Premium |
Claude Opus 4 |
$15 input |
*Hardware costs not included
Each model card follows a consistent structure:
# Model Name
## Overview
- Organization, release date, key claims
## Architecture
- Parameters, layers, context, special features
## Benchmarks
- Performance on standard evals
## Usage
- API access, pricing, integrations
## Comparison
- How it stacks up vs alternatives
## Resources
- Links to papers, docs, demos
- Create
cards/<model-name>.md using the template
- Update this README's index
- Add any PDF/raw model cards to
raw/
- Submit a PR!
- Fix outdated pricing or specs
- Add new comparisons or benchmarks
- Improve explanations
- Add code examples
model-cards-hub/
├── README.md # You are here
├── cards/ # Individual model cards
│ ├── _TEMPLATE.md
│ ├── claude-opus-4.md
│ ├── deepseek-r1.md
│ ├── kimi-k2.md
│ └── ...
├── guides/ # Reference guides
│ ├── comparison-matrix.md
│ ├── pricing-comparison.md
│ ├── model-selection-guide.md
│ ├── api-snippets.md
│ ├── self-hosting-guide.md
│ ├── license-comparison.md
│ ├── model-family-trees.md
│ ├── models-to-watch.md
│ └── glossary.md
├── raw/ # Original PDFs and docs
└── notes/ # Working notes
Maintained by codmire_'s assistant 🐈⬛
Last updated: February 2026