---
title: "AI models released in January 2025"
type: "release-archive"
canonical_url: "https://theaiengineering.com/releases/2025/01"
markdown_url: "https://theaiengineering.com/releases/2025/01.md"
author: "The AI Engineering Editors"
publisher: "The AI Engineering"
updated: "2025-01-31"
citation: "The AI Engineering Editors. \"AI models released in January 2025.\" The AI Engineering, updated Jan 31, 2025. https://theaiengineering.com/releases/2025/01"
license: "CC BY 4.0"
license_url: "https://creativecommons.org/licenses/by/4.0/"
data_csv: "https://theaiengineering.com/releases.csv"
data_json: "https://theaiengineering.com/releases.json"
---

# AI models released in January 2025

25 releases from 15 labs, 13 with open weights. Dates are the labs' official announcement dates.

## January 2025

### OpenAI o3-mini

o3-mini is a small, cost-efficient reasoning model optimized for STEM tasks such as science, math, and coding. It replaced o1-mini in ChatGPT and offers low, medium, and high reasoning effort in the API.

- Date: 2025-01-31
- Lab: OpenAI
- Type: Reasoning
- Weights: Proprietary
- Modalities: text → text
- Context window: 200,000 tokens
- Available at launch: Consumer app, API
- Permalink: https://theaiengineering.com/releases/2025/01#o3-mini

Highlights:

- API model ID o3-mini (snapshot o3-mini-2025-01-31)
- First OpenAI small reasoning model with function calling, Structured Outputs, and developer messages
- Free ChatGPT users could try it via 'Reason'; paid users also got o3-mini-high
- No vision support; 200,000-token context window per the API model page

Source: [OpenAI o3-mini | OpenAI](https://openai.com/index/openai-o3-mini/)

### Mistral Small 3

Latency-optimized 24B model released under Apache 2.0 with pretrained and instruct checkpoints, marking Mistral's return to Apache licensing for general-purpose models.

- Date: 2025-01-30
- Lab: Mistral AI
- Type: LLM
- Weights: Open weights
- Modalities: text → text
- Context window: 32,000 tokens
- Parameters: 24B
- License: Apache-2.0
- Available at launch: Open weights, API
- Permalink: https://theaiengineering.com/releases/2025/01#mistral-small-3

Highlights:

- Uses far fewer layers than comparable models to cut latency
- Not trained with RL or synthetic data
- API model ID mistral-small-2501

Source: [Mistral Small 3 | Mistral AI](https://mistral.ai/news/mistral-small-3)

### SFR-Embedding-Code

Family of embedding models for code and text retrieval in 400M, 2B and 7B sizes, with the 400M and 2B models released openly.

- Date: 2025-01-30
- Lab: Salesforce
- Type: Embedding
- Weights: Open weights
- Modalities: text → embedding
- Parameters: 400M and 2B
- License: CC-BY-NC-4.0
- Available at launch: Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#sfr-embedding-code

Highlights:

- The 7B model ranked first on the CoIR code retrieval benchmark at announcement
- The 400M model was designed for faster, lower-cost inference, including on CPUs
- Open models released for research use under CC BY-NC 4.0

Source: [SFR-Embedding-Code: A Family of Embedding Models for Code Retrieval](https://www.salesforce.com/blog/sfr-embedding-code/)

### Tülu 3 405B

Application of the fully open Tülu 3 post-training recipe to the Llama 3.1 405B base model, used to stress-test reinforcement learning with verifiable rewards at scale.

- Date: 2025-01-30
- Lab: Ai2
- Type: LLM
- Weights: Open weights
- Modalities: text → text
- Parameters: 405B
- License: Llama 3.1 Community License
- Available at launch: Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#tulu-3-405b

Highlights:

- Trained with reinforcement learning from verifiable rewards
- Served with vLLM using 16-way tensor parallelism during RL
- Built on Llama 3.1 405B

Source: [Scaling the Tülu 3 post-training recipes to surpass the performance of DeepSeek V3 | Ai2](https://allenai.org/blog/tulu-3-405b)

### Qwen2.5-Max

Qwen2.5-Max is a large-scale mixture-of-experts model pre-trained on more than 20T tokens and post-trained with supervised fine-tuning and RLHF. Qwen released its API through Alibaba Cloud.

- Date: 2025-01-28
- Lab: Alibaba Qwen
- Type: LLM
- Weights: Proprietary
- Modalities: text → text
- Available at launch: API, Consumer app
- Permalink: https://theaiengineering.com/releases/2025/01#qwen2-5-max

Highlights:

- Mixture-of-experts architecture
- API available through Alibaba Cloud Model Studio
- Available to try on Qwen Chat

Source: [Qwen2.5-Max: Exploring the Intelligence of Large-scale MoE Model](https://qwenlm.github.io/blog/qwen2.5-max/)

### Janus-Pro

Janus-Pro is an advanced version of Janus, DeepSeek's autoregressive framework that unifies multimodal understanding and image generation by decoupling visual encoding into separate pathways. It significantly improves both understanding and visual generation.

- Date: 2025-01-27
- Lab: DeepSeek
- Type: Multimodal
- Weights: Open weights
- Modalities: text, image → text, image
- Parameters: 1B, 7B
- License: DeepSeek Model License
- Available at launch: Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#janus-pro

Highlights:

- Built on DeepSeek-LLM-1.5b-base and DeepSeek-LLM-7b-base
- Single unified transformer for image understanding and text-to-image generation
- Code under the MIT License; model use under the DeepSeek Model License

Source: [Janus-Series: Unified Multimodal Understanding and Generation Models (GitHub)](https://github.com/deepseek-ai/Janus)

### Qwen2.5-1M

Qwen2.5-1M is the first set of open Qwen models upgraded to handle 1M-token contexts, released as Qwen2.5-7B-Instruct-1M and Qwen2.5-14B-Instruct-1M with a vLLM-based inference framework.

- Date: 2025-01-27
- Lab: Alibaba Qwen
- Type: LLM
- Weights: Open weights
- Modalities: text → text
- Context window: 1,010,000 tokens
- Parameters: 7B, 14B
- License: Apache-2.0
- Available at launch: Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#qwen2-5-1m

Highlights:

- Context length of 1,010,000 tokens with 8,192-token generation
- Inference framework with sparse attention processes 1M-token inputs 3 to 7 times faster
- Weights under the Apache 2.0 license

Source: [Qwen2.5-1M: Deploy Your Own Qwen with Context Length up to 1M Tokens](https://qwenlm.github.io/blog/qwen2.5-1m/)

### Qwen2.5-VL

Qwen2.5-VL is Qwen's vision-language model series released in 3B, 7B, and 72B sizes as base and instruct models. The flagship Qwen2.5-VL-72B-Instruct covers document understanding, video understanding, and visual agent tasks.

- Date: 2025-01-26
- Lab: Alibaba Qwen
- Type: Multimodal
- Weights: Open weights
- Modalities: text, image, video → text
- Context window: 32,768 tokens
- Parameters: 3B, 7B, 72B
- License: Apache-2.0 (7B); Qwen License Agreement (72B); Qwen Research License Agreement (3B)
- Available at launch: Consumer app, Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#qwen2-5-vl

Highlights:

- Qwen2.5-VL-72B-Instruct available on Qwen Chat
- Qwen2.5-VL-3B targets edge AI
- Weights on Hugging Face and ModelScope

Source: [Qwen2.5 VL! Qwen2.5 VL! Qwen2.5 VL!](https://qwenlm.github.io/blog/qwen2.5-vl/)

### Hunyuan-7B-0124

Hunyuan-7B-0124 is an open 7B dense language model released in pretrained and instruct versions, using improved data allocation and training, grouped query attention, and a context window extended to 256K.

- Date: 2025-01-24
- Lab: Tencent
- Type: LLM
- Weights: Open weights
- Modalities: text → text
- Context window: 256,000 tokens
- Parameters: 7B
- License: Tencent Hunyuan Community License Agreement
- Available at launch: Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#hunyuan-7b-0124

Highlights:

- Released as Hunyuan-7B-Pretrain-0124 and Hunyuan-7B-Instruct-0124
- Hugging Face format with a vLLM inference backend

Source: [GitHub - Tencent-Hunyuan/Tencent-Hunyuan-7B-0124](https://github.com/Tencent-Hunyuan/Tencent-Hunyuan-7B-0124)

### Computer-Using Agent (computer-use-preview)

Computer-Using Agent (CUA) combines GPT-4o's vision with reasoning trained through reinforcement learning to operate graphical interfaces from screenshots using a virtual mouse and keyboard. It launched powering the Operator research preview.

- Date: 2025-01-23
- Lab: OpenAI
- Type: Multimodal
- Weights: Proprietary
- Modalities: text, image → text
- Context window: 8,192 tokens
- Available at launch: Consumer app
- Permalink: https://theaiengineering.com/releases/2025/01#computer-using-agent

Highlights:

- Launched in Operator for ChatGPT Pro users in the U.S.
- Released in the API as computer-use-preview with the Responses API on March 11, 2025
- Asks for user confirmation on sensitive actions such as logins and CAPTCHAs
- API model page lists an 8,192-token context window

Source: [Computer-Using Agent | OpenAI](https://openai.com/index/computer-using-agent/)

### SmolVLM 256M and 500M

Smaller additions to the SmolVLM family, with the 256M model presented as the smallest vision-language model in the world at release.

- Date: 2025-01-23
- Lab: Hugging Face
- Type: Multimodal
- Weights: Open weights
- Modalities: text, image → text
- Parameters: 256M and 500M
- License: Apache-2.0
- Available at launch: Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#smolvlm-256m-500m

Highlights:

- Released with ColSmolVLM retrieval models
- Base and instruct versions
- Released under Apache 2.0

Source: [SmolVLM Grows Smaller – Introducing the 256M & 500M Models!](https://huggingface.co/blog/smolervlm)

### Doubao-1.5-pro

Doubao-1.5-pro is ByteDance's sparse mixture-of-experts flagship model, built with an integrated training and inference design that uses few active parameters, with upgraded vision and end-to-end speech capabilities.

- Date: 2025-01-22
- Lab: ByteDance Seed
- Type: LLM
- Weights: Proprietary
- Modalities: text, image → text
- Context window: 256,000 tokens
- Available at launch: API
- Permalink: https://theaiengineering.com/releases/2025/01#doubao-1-5-pro

Highlights:

- Offered on Volcano Engine in 32K and 256K context versions
- Released alongside Doubao-1.5-lite and Doubao-1.5-vision-pro
- Uses a 2.4B-parameter Doubao ViT with native dynamic resolution
- Post-training used no data from other models

Source: [Doubao 1.5pro - Doubao Team](https://seed.bytedance.com/en/special/doubao_1_5_pro)

### Hunyuan3D 2.0

Hunyuan3D 2.0 is an open large-scale 3D asset generation system that pairs a shape generation model with a texture synthesis model to produce high-resolution textured 3D assets from images.

- Date: 2025-01-21
- Lab: Tencent
- Type: Image
- Weights: Open weights
- Modalities: image → 3d
- License: Tencent Hunyuan 3D 2.0 Community License Agreement
- Available at launch: Open weights, Consumer app
- Permalink: https://theaiengineering.com/releases/2025/01#hunyuan3d-2-0

Highlights:

- Hunyuan3D-DiT-v2-0 shape model has 1.1B parameters and Hunyuan3D-Paint-v2-0 has 1.3B
- Launched alongside the Hunyuan3D Studio website

Source: [GitHub - Tencent-Hunyuan/Hunyuan3D-2: High-Resolution 3D Assets Generation with Large Scale Hunyuan3D Diffusion Models.](https://github.com/Tencent-Hunyuan/Hunyuan3D-2)

### Sonar and Sonar Pro (API)

Perplexity's Sonar API models with built-in real-time web search and citations: Sonar for fast, lightweight answers and Sonar Pro for multi-step queries with more citations and a larger context window.

- Date: 2025-01-21
- Lab: Perplexity
- Type: LLM
- Weights: Proprietary
- Modalities: text → text
- Available at launch: API
- Permalink: https://theaiengineering.com/releases/2025/01#perplexity-sonar-pro

Highlights:

- API model IDs sonar and sonar-pro; Sonar Pro generally available to all developers at launch
- Model docs listed 127k context for sonar and 200k for sonar-pro, with an 8k output cap on sonar-pro
- Replaced the llama-3.1-sonar online models, which were retired on February 22, 2025
- JSON mode and search domain filters opened to select usage tiers

Source: [Introducing the Sonar Pro API by Perplexity](https://www.perplexity.ai/hub/blog/introducing-the-sonar-pro-api)

### DeepSeek-R1

DeepSeek-R1 is an open-weights reasoning model trained with large-scale reinforcement learning in post-training, released with a technical report. DeepSeek also released DeepSeek-R1-Zero, trained with reinforcement learning without supervised fine-tuning.

- Date: 2025-01-20
- Lab: DeepSeek
- Type: Reasoning
- Weights: Open weights
- Modalities: text → text
- Context window: 128,000 tokens
- Parameters: MoE: 671B total, 37B active
- License: MIT
- Available at launch: API, Consumer app, Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#deepseek-r1

Highlights:

- Code and weights under the MIT License; API outputs may be used for fine-tuning and distillation
- Available in the API as deepseek-reasoner and as DeepThink on chat.deepseek.com
- 128K context length per the model card

Source: [DeepSeek-R1 Release](https://api-docs.deepseek.com/news/news250120)

### DeepSeek-R1-Distill

DeepSeek-R1-Distill is a set of six dense models fine-tuned on reasoning samples generated by DeepSeek-R1, based on Qwen2.5 and Llama 3 checkpoints. They bring R1-style reasoning to sizes from 1.5B to 70B parameters.

- Date: 2025-01-20
- Lab: DeepSeek
- Type: Reasoning
- Weights: Open weights
- Modalities: text → text
- Parameters: 1.5B, 7B, 8B, 14B, 32B, 70B
- License: MIT
- Available at launch: Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#deepseek-r1-distill

Highlights:

- Fine-tuned with 800K samples curated with DeepSeek-R1
- Qwen-based models derive from Qwen2.5 (Apache 2.0); Llama-based models from Llama 3.1 8B and Llama 3.3 70B
- Released the same day as DeepSeek-R1

Source: [DeepSeek-R1 Release](https://api-docs.deepseek.com/news/news250120)

### Doubao Realtime Voice Model

The Doubao Realtime Voice Model is an end-to-end speech understanding and generation model that replaces cascaded pipelines for spoken conversation, offering expressive voice control, emotional responsiveness, low latency, and smooth interruption.

- Date: 2025-01-20
- Lab: ByteDance Seed
- Type: Speech
- Weights: Proprietary
- Modalities: audio → audio
- Available at launch: Consumer app
- Permalink: https://theaiengineering.com/releases/2025/01#doubao-realtime-voice

Highlights:

- Fully available in the Doubao app from version 7.2.0
- Integrates speech and text modalities natively
- Supports natural interruptions during dialogue

Source: [Doubao Realtime Voice Model Is Available Upon Release! High EQ and IQ](https://seed.bytedance.com/en/blog/doubao-realtime-voice-model-is-available-upon-release-high-eq-and-iq)

### Kimi k1.5

Kimi k1.5 is Moonshot AI's multimodal reasoning model trained with reinforcement learning, which Moonshot describes as o1-level. Its training scaled the reinforcement learning context window to 128K and used partial rollouts for efficiency.

- Date: 2025-01-20
- Lab: Moonshot AI
- Type: Reasoning
- Weights: Proprietary
- Modalities: text, image → text
- Available at launch: Consumer app
- Permalink: https://theaiengineering.com/releases/2025/01#kimi-k1-5

Highlights:

- Jointly trained on text and vision data
- Reports long chain-of-thought and short chain-of-thought results
- Moonshot said the service would come to kimi.ai soon after the January 20, 2025 report

Source: [Kimi k1.5: Scaling Reinforcement Learning with LLMs (GitHub)](https://github.com/MoonshotAI/Kimi-k1.5)

### LFM-7B

A 7.7B-parameter Liquid Foundation Model tuned for chat, instruction following, and multilingual use, with Liquid AI highlighting English, Arabic, and Japanese, aimed at private enterprise, on-premises, and on-device deployment.

- Date: 2025-01-20
- Lab: Liquid AI
- Type: LLM
- Weights: Proprietary
- Modalities: text → text
- Parameters: 7.7B
- Available at launch: API, Consumer app, Cloud platforms, Enterprise
- Permalink: https://theaiengineering.com/releases/2025/01#lfm-7b

Highlights:

- Supports English, Spanish, French, German, Chinese, Arabic, Japanese, and Korean
- Effective context length of 32k tokens validated with RULER
- Offered on Liquid Playground, OpenRouter, and AWS Marketplace
- Weights not released; on-premises customization stack sold to enterprises

Source: [Introducing LFM-7B: Setting New Standards for Efficient Language Models](https://www.liquid.ai/blog/introducing-lfm-7b-setting-new-standards-for-efficient-language-models)

### MiniMax-Text-01 and MiniMax-VL-01

MiniMax-01 is an open-weight series pairing the MiniMax-Text-01 language model with the MiniMax-VL-01 vision-language model, built on a hybrid of Lightning Attention, softmax attention, and mixture-of-experts.

- Date: 2025-01-15
- Lab: MiniMax
- Type: LLM
- Weights: Open weights
- Modalities: text, image → text
- Context window: 4,000,000 tokens
- Parameters: MoE: 456B total, 45.9B active
- License: MiniMax Model License
- Available at launch: API, Consumer app, Open weights
- Permalink: https://theaiengineering.com/releases/2025/01#minimax-01

Highlights:

- 456B total parameters with 45.9B activated per token
- Trained at a 1M-token context and handles up to 4M tokens at inference
- VL-01 adds a 303M-parameter vision transformer to the Text-01 base
- Model weights under the MiniMax Model License; code under MIT

Source: [MiniMax-01 is Now Open-Source: Scaling Lightning Attention for the AI Agent Era](https://www.minimax.io/news/minimax-01-series-2)

### Ray2

Second-generation Luma video model built on a new multimodal architecture trained with ten times the compute of Ray1, available in Dream Machine and later with learnable camera motion concepts.

- Date: 2025-01-15
- Lab: Luma AI
- Type: Video
- Weights: Proprietary
- Modalities: text, image → video
- Available at launch: Consumer app
- Permalink: https://theaiengineering.com/releases/2025/01#luma-ray2

Highlights:

- Ten times the compute of Ray1
- Camera Motion Concepts added March 31, 2025
- Announced for Adobe Firefly in April 2025

Source: [Changelog | Luma](https://lumalabs.ai/changelog)

### Codestral 25.01

Update of Codestral with a more efficient architecture and improved tokenizer that generates and completes code about twice as fast as the original, aimed at fill-in-the-middle use.

- Date: 2025-01-13
- Lab: Mistral AI
- Type: Coding
- Weights: Proprietary
- Modalities: text → text
- Available at launch: API
- Permalink: https://theaiengineering.com/releases/2025/01#codestral-2501

Highlights:

- Supports fill-in-the-middle, code correction and test generation
- Optimized for low-latency, high-frequency coding tasks
- API model ID codestral-2501

Source: [Codestral 25.01 | Mistral AI](https://mistral.ai/news/codestral-2501)

### MiniMax I2V-01-live

I2V-01-live is a Hailuo image-to-video model in the I2V family designed to animate 2D illustrations, with smoother, more vivid motion, greater stability, and subtle expressions across a range of art styles.

- Date: 2025-01-12
- Lab: MiniMax
- Type: Video
- Weights: Proprietary
- Modalities: text, image → video
- Available at launch: API, Consumer app
- Permalink: https://theaiengineering.com/releases/2025/01#minimax-i2v-01-live

Highlights:

- Aimed at bringing 2D illustrations and characters to life
- Available in Hailuo AI and through the MiniMax API

Source: [I2V-01-live: smooth, vivid, and perfect for 2D illustrations](https://www.minimax.io/news/i2v-01-live-release)

### MiniMax S2V-01

S2V-01 is Hailuo's subject reference video model, generating videos that keep a person's facial features and identity consistent across camera angles and movement from a single reference image.

- Date: 2025-01-10
- Lab: MiniMax
- Type: Video
- Weights: Proprietary
- Modalities: text, image → video
- Available at launch: API, Consumer app
- Permalink: https://theaiengineering.com/releases/2025/01#minimax-s2v-01

Highlights:

- Needs only one reference image
- Posture, expression, and lighting can be adjusted with text prompts

Source: [S2V-01: Subject Reference of Hailuo](https://www.minimax.io/news/s2v-01-release)

### Cosmos World Foundation Models

Platform of generative world foundation models for physical AI, with diffusion and autoregressive models, tokenizers and guardrails for robotics and autonomous vehicles.

- Date: 2025-01-06
- Lab: NVIDIA
- Type: Video
- Weights: Open weights
- Modalities: text, image, video → video
- License: NVIDIA Open Model License
- Available at launch: Open weights, API
- Permalink: https://theaiengineering.com/releases/2025/01#cosmos-1-0

Highlights:

- Announced at CES 2025
- Weights on Hugging Face and NVIDIA NGC under NVIDIA's open model license
- Previewable in the NVIDIA API catalog

Source: [NVIDIA Launches Cosmos World Foundation Model Platform to Accelerate Physical AI Development | NVIDIA Newsroom](https://nvidianews.nvidia.com/news/nvidia-launches-cosmos-world-foundation-model-platform-to-accelerate-physical-ai-development)

## Cite and reuse this data

These release records are free to reuse under CC BY 4.0, including commercially. Credit The AI Engineering and link to this page's URL. Dates are the labs' official announcement dates, and every record links to its source. List fields in the CSV are separated by " | ".

The AI Engineering Editors. "AI models released in January 2025." The AI Engineering, updated Jan 31, 2025. https://theaiengineering.com/releases/2025/01

HTML link:

```html
<a href="https://theaiengineering.com/releases/2025/01">AI models released in January 2025</a> (The AI Engineering)
```

- Title: AI models released in January 2025
- Author: The AI Engineering Editors
- Publisher: The AI Engineering
- Updated: 2025-01-31
- URL: https://theaiengineering.com/releases/2025/01
- Markdown: https://theaiengineering.com/releases/2025/01.md
- License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)
- Full dataset as CSV: https://theaiengineering.com/releases.csv
- Full dataset as JSON: https://theaiengineering.com/releases.json
