分析新AI模型的发布速度。按服务商追踪发布节奏，识别加速趋势，查看哪些服务商正在发布最多的新模型。
325
有发布日期
27
个模型发布
50
个模型发布
Reka Edge
5 days ago
最近 12 个月的月度模型发布。平均每月约 9.5 个模型。
| 月份 | 数量 |
|---|---|
| Mar 2026 | 18 |
| Feb 2026 | 19 |
| Jan 2026 | 13 |
| Dec 2025 | 26 |
| Nov 2025 | 11 |
| Oct 2025 | 25 |
| Sep 2025 | 22 |
| Aug 2025 | 18 |
| Jul 2025 | 14 |
| Jun 2025 | 12 |
| May 2025 | 11 |
| Apr 2025 | 21 |
总产量最多: OpenAI。近期最活跃（90天）: OpenAI。显示拥有2+模型的服务商。
| 提供商 | 总计 | 30d | 90d |
|---|---|---|---|
| OpenAI | 64 | 6 | 9 |
| Alibaba | 50 | 5 | 9 |
| 28 | 3 | 4 | |
| Mistral AI | 26 | 1 | 1 |
| Meta | 14 | — | — |
| Anthropic | 13 | — | 2 |
| NVIDIA | 11 | 2 | 2 |
| DeepSeek | 11 | — | — |
| xAI | 10 | 2 | 2 |
| MiniMax | 9 | 1 | 4 |
| arcee-ai | 7 | — | 1 |
| ByteDance | 5 | 2 | 2 |
| Liquid AI | 5 | 1 | 3 |
| Amazon | 5 | — | — |
| Perplexity | 5 | — | — |
| Baidu | 5 | — | — |
| Moonshot AI | 4 | — | 1 |
| Allen AI | 4 | — | 1 |
| Cohere | 4 | — | — |
| Xiaomi | 3 | 2 | 2 |
| Inception | 3 | 1 | 1 |
| aion-labs | 3 | — | 1 |
| StepFun | 2 | — | 2 |
| Cursor | 2 | — | — |
| Kuaishou | 2 | — | — |
| Microsoft | 2 | — | — |
| Stability AI | 2 | — | — |
| Inflection | 2 | — | — |
27 个模型在最近30天内发布,按发布日期排序。
| 模型 | 提供商 | 发布于 |
|---|---|---|
| Reka Edge | reka | Mar 20, 2026 |
| MiMo-V2-Omni | Xiaomi | Mar 18, 2026 |
| MiMo-V2-Pro | Xiaomi | Mar 18, 2026 |
| MiniMax M2.7 | MiniMax | Mar 18, 2026 |
| GPT-5.4 Nano | OpenAI | Mar 17, 2026 |
| GPT-5.4 Mini | OpenAI | Mar 17, 2026 |
| Mistral Small 4 | Mistral AI | Mar 16, 2026 |
| Grok 4.20 Multi-Agent Beta | xAI | Mar 12, 2026 |
| Grok 4.20 Beta | xAI | Mar 12, 2026 |
| Nemotron 3 Super (free) | NVIDIA | Mar 11, 2026 |
| Nemotron 3 Super | NVIDIA | Mar 11, 2026 |
| Seed-2.0-Lite | ByteDance | Mar 10, 2026 |
| Qwen3.5-9B | Alibaba | Mar 10, 2026 |
| GPT-5.4 Pro | OpenAI | Mar 5, 2026 |
| GPT-5.4 | OpenAI | Mar 5, 2026 |
| Mercury 2 | Inception | Mar 4, 2026 |
| GPT-5.3 Chat | OpenAI | Mar 3, 2026 |
| Gemini 3.1 Flash Lite Preview | Mar 3, 2026 | |
| Seed-2.0-Mini | ByteDance | Feb 26, 2026 |
| Nano Banana 2 (Gemini 3.1 Flash Image Preview) | Feb 26, 2026 | |
| Qwen3.5-35B-A3B | Alibaba | Feb 25, 2026 |
| Qwen3.5-27B | Alibaba | Feb 25, 2026 |
| Qwen3.5-122B-A10B | Alibaba | Feb 25, 2026 |
| Qwen3.5-Flash | Alibaba | Feb 25, 2026 |
| LFM2-24B-A2B | Liquid AI | Feb 25, 2026 |
| Gemini 3.1 Pro Preview Custom Tools | Feb 25, 2026 | |
| GPT-5.3-Codex | OpenAI | Feb 24, 2026 |
比较每个服务商的近期发布速率(最近90天)与总体速率。近期发布更多模型的服务商正在加速。
| 提供商 | 总体速率/月 | 近期速率/月 | 趋势 |
|---|---|---|---|
| OpenAI | 1.9 | 3.0 | 加速中 |
| Alibaba | 2.8 | 3.0 | 稳定 |
| 1.3 | 1.3 | 稳定 | |
| Mistral AI | 0.9 | 0.3 | 减速中 |
| Meta | 1.1 | 0.0 | 减速中 |
| Anthropic | 0.6 | 0.7 | 加速中 |
| NVIDIA | 0.6 | 0.7 | 稳定 |
| DeepSeek | 1.0 | 0.0 | 减速中 |
| xAI | 0.9 | 0.7 | 减速中 |
| MiniMax | 0.5 | 1.3 | 加速中 |
| arcee-ai | 0.8 | 0.3 | 减速中 |
| ByteDance | 0.6 | 0.7 | 稳定 |
| Liquid AI | 1.2 | 1.0 | 减速中 |
| Amazon | 0.4 | 0.0 | 减速中 |
| Perplexity | 0.5 | 0.0 | 减速中 |
| Baidu | 1.5 | 0.0 | 减速中 |
| Moonshot AI | 0.6 | 0.3 | 减速中 |
| Allen AI | 0.4 | 0.3 | 减速中 |
| Cohere | 0.6 | 0.0 | 减速中 |
| Xiaomi | 1.0 | 0.7 | 减速中 |
| Inception | 0.3 | 0.3 | 加速中 |
| aion-labs | 0.2 | 0.3 | 加速中 |
| StepFun | 2.0 | 0.7 | 减速中 |
| Cursor | 2.0 | 0.0 | 减速中 |
| Kuaishou | 0.1 | 0.0 | 减速中 |
| Microsoft | 0.2 | 0.0 | 减速中 |
| Stability AI | 0.2 | 0.0 | 减速中 |
| Inflection | 2.0 | 0.0 | 减速中 |
当前追踪的模型有多久了?按年龄段分类。
Release velocity measures how frequently AI providers release new models. It tracks the pace of new model launches over time, helping identify which providers are most actively developing and which periods see the most activity.
The release timeline above shows monthly model counts. The AI industry has been accelerating, with some providers releasing multiple models per month. Check the Release Overview section for exact counts over the last 30, 90, and 180 days.
The most prolific provider is shown in the Provider Release Activity section. Some providers focus on releasing many model variants, while others concentrate on fewer but more impactful releases.
Acceleration compares a provider's recent release rate (last 90 days) to their overall historical rate. Providers shipping more models recently than their average are classified as Accelerating, while those slowing down are Decelerating.