Track the latest AI model releases and LLM updates. 24 new models have been released in the last 30 days across 10 providers. Scores, pricing, and capabilities are updated hourly.
| Model | Score |
|---|---|
| Grok 4.20 Multi-Agent BetaxAI | 81 |
| Grok 4.20 BetaxAI | 88 |
| Nemotron 3 Super (free)NVIDIA | 88 |
| Seed-2.0-LiteByteDance | 88 |
| Qwen3.5-9BAlibaba | 78 |
| Model | Score |
|---|---|
| GPT-5.4 ProOpenAI | 97 |
| GPT-5.4OpenAI | 97 |
| Mercury 2Inception | 78 |
| GPT-5.3 ChatOpenAI | 84 |
| Gemini 3.1 Flash Lite PreviewGoogle | 89 |
| Model | Score |
|---|---|
| Seed-2.0-MiniByteDance | 88 |
| Nano Banana 2 (Gemini 3.1 Flash Image Preview)Google | — |
| Qwen3.5-35B-A3BAlibaba | 87 |
| Qwen3.5-27BAlibaba | 87 |
| Qwen3.5-122B-A10BAlibaba | 87 |
| Qwen3.5-FlashAlibaba | 89 |
| LFM2-24B-A2BLiquid AI | 45 |
| Gemini 3.1 Pro Preview Custom ToolsGoogle | 89 |
| GPT-5.3-CodexOpenAI | 96 |
| Aion-2.0aion-labs | 63 |
| Gemini 3.1 Pro PreviewGoogle | 89 |
| Claude Sonnet 4.6Anthropic | 97 |
| Qwen3.5 Plus 2026-02-15Alibaba | 89 |
| Qwen3.5 397B A17BAlibaba | 87 |
The AI industry is releasing new models at an unprecedented pace. Major providers like OpenAI, Anthropic, Google, and Meta ship updates weekly, while open-source models from DeepSeek, Qwen, and Mistral often appear within days of each other. Tracking these releases helps developers choose the best model for their workloads.
Every new model is automatically scored using our composite methodology that weighs capabilities (25%), pricing (25%), context window (15%), recency (15%), output capacity (10%), and versatility (10%). Scores update hourly as we gather more benchmark data and community feedback.
The pace of AI model releases has accelerated significantly. In the past 30 days alone, 24 new models were released across 10 providers. We track every new model as it becomes available through API providers.
We monitor API providers including OpenRouter, OpenAI, Anthropic, Google, and others. Our system checks for new models hourly and automatically scores them using our composite ranking methodology that evaluates capabilities, pricing, context window, and more.
Each model receives a composite score from 0 to 100, calculated from six signals: capabilities (25%), pricing tier (25%), context window (15%), recency (15%), output capacity (10%), and versatility (10%). Higher scores indicate a stronger overall offering.
OpenAI, Google, Anthropic, Meta, and DeepSeek are among the most active providers. Open-source model releases from companies like Meta (Llama), Mistral, and Qwen have also increased significantly, often appearing within days of each other.
This page is updated hourly with the latest model releases. You can also check our model release tracker for a timeline view, or visit the full leaderboard to see how new models rank against existing ones.