AI model releases | Interestana AI Index

Mistral Large 3 (Mistral AI)

Tue, 02 Dec 2025 00:00:00 GMT

Mistral AI released Mistral Large 3 on December 2, 2025.

Mistral Large 3 is an open-weight, sparse mixture-of-experts model from Mistral AI, released on December 2, 2025 under Apache 2.0, with 41B active and 675B total parameters. It accepts text and image input.

Mistral AI's page

Claude Haiku 4.5 (Anthropic)

Wed, 15 Oct 2025 00:00:00 GMT

Anthropic released Claude Haiku 4.5 on October 15, 2025.

Claude Haiku 4.5 is a model in Anthropic’s Haiku line, released on October 15, 2025 (API ID claude-haiku-4-5-20251001). It takes text and image input, returns text, and supports extended thinking.

Anthropic's page

Claude Sonnet 4.5 (Anthropic)

Mon, 29 Sep 2025 00:00:00 GMT

Anthropic released Claude Sonnet 4.5 on September 29, 2025.

Claude Sonnet 4.5 is a model from Anthropic, released on September 29, 2025 (API ID claude-sonnet-4-5-20250929). It takes text and image input, returns text, and supports extended thinking.

Anthropic's page

GPT-5 (OpenAI)

Thu, 07 Aug 2025 00:00:00 GMT

OpenAI released GPT-5 on August 7, 2025.

GPT-5 is a reasoning model from OpenAI, released in the API on August 7, 2025 (snapshot gpt-5-2025-08-07). It takes text and image input, returns text, and has configurable reasoning effort.

OpenAI's page

Gemini 2.5 Flash-Lite (Google DeepMind)

Tue, 22 Jul 2025 00:00:00 GMT

Google DeepMind released Gemini 2.5 Flash-Lite on July 22, 2025.

Gemini 2.5 Flash-Lite is a lower-priced Gemini model from Google, released in preview on June 17, 2025 and as a stable model (gemini-2.5-flash-lite) on July 22, 2025. It takes text, image, video, audio and PDF input and returns text.

Google DeepMind's page

Grok 4 (xAI)

Wed, 09 Jul 2025 00:00:00 GMT

xAI released Grok 4 on July 9, 2025.

Grok 4 is a model from xAI, launched on July 9, 2025 alongside Grok 4 Heavy. xAI says it was trained with reinforcement learning to use tools, including real-time search, and that its API handles text and vision.

xAI's page

Gemini 2.5 Pro (Google DeepMind)

Tue, 17 Jun 2025 00:00:00 GMT

Google DeepMind released Gemini 2.5 Pro on June 17, 2025.

Gemini 2.5 Pro is a model from Google, first released as an experimental model on March 25, 2025 and as a stable model (gemini-2.5-pro) on June 17, 2025. It takes text, image, audio, video and PDF input and returns text.

Google DeepMind's page

Claude Opus 4 (Anthropic)

Thu, 22 May 2025 00:00:00 GMT

Anthropic released Claude Opus 4 on May 22, 2025.

Claude Opus 4 is a model from Anthropic, released on May 22, 2025 alongside Claude Sonnet 4, with both near-instant and extended-thinking modes.

Anthropic's page

Qwen3-235B-A22B (Alibaba)

Tue, 29 Apr 2025 00:00:00 GMT

Alibaba released Qwen3-235B-A22B on April 29, 2025.

Qwen3-235B-A22B is an open-weight mixture-of-experts model in Alibaba’s Qwen3 family, released on April 29, 2025 under Apache 2.0, with 235B total and 22B activated parameters.

Alibaba's page

OpenAI o3 (OpenAI)

Wed, 16 Apr 2025 00:00:00 GMT

OpenAI released OpenAI o3 on April 16, 2025.

o3 is a reasoning model from OpenAI, released on April 16, 2025 (snapshot o3-2025-04-16). It takes text and image input and returns text.

OpenAI's page

Llama 4 Maverick (Meta)

Sat, 05 Apr 2025 00:00:00 GMT

Meta released Llama 4 Maverick on April 5, 2025.

Llama 4 Maverick is an open-weight mixture-of-experts model from Meta, released on April 5, 2025, with 17B active and 400B total parameters. It takes multilingual text and image input and returns text and code.

Meta's page

Llama 4 Scout (Meta)

Sat, 05 Apr 2025 00:00:00 GMT

Meta released Llama 4 Scout on April 5, 2025.

Llama 4 Scout is an open-weight mixture-of-experts model from Meta, released on April 5, 2025, with 17B active and 109B total parameters. It takes multilingual text and image input and returns text and code.

Meta's page

Gemini 2.0 Flash (Google DeepMind)

Wed, 05 Feb 2025 00:00:00 GMT

Google DeepMind released Gemini 2.0 Flash on February 5, 2025.

Gemini 2.0 Flash is a model from Google, made generally available as gemini-2.0-flash-001 on February 5, 2025, with multimodal input and text output.

Google DeepMind's page

DeepSeek-R1 (DeepSeek)

Mon, 20 Jan 2025 00:00:00 GMT

DeepSeek released DeepSeek-R1 on January 20, 2025.

DeepSeek-R1 is an open-weight reasoning model from DeepSeek, released on January 20, 2025 under the MIT licence, with 671B total and 37B activated parameters.

DeepSeek's page

DeepSeek-V3 (DeepSeek)

Thu, 26 Dec 2024 00:00:00 GMT

DeepSeek released DeepSeek-V3 on December 26, 2024.

DeepSeek-V3 is an open-weight mixture-of-experts language model from DeepSeek, announced on December 26, 2024, with 671B total and 37B activated parameters, trained on 14.8T tokens.

DeepSeek's page

Phi-4 (Microsoft Research)

Thu, 12 Dec 2024 00:00:00 GMT

Microsoft Research released Phi-4 on December 12, 2024.

Phi-4 is a 14B-parameter open-weight language model from Microsoft Research, released on December 12, 2024 under the MIT licence. It takes text input and is intended mainly for English.

Microsoft Research's page

Amazon Nova Pro (Amazon)

Tue, 03 Dec 2024 00:00:00 GMT

Amazon released Amazon Nova Pro on December 3, 2024.

Amazon Nova Pro is a multimodal model from Amazon, announced on December 3, 2024 and offered through Amazon Bedrock (model ID amazon.nova-pro-v1:0). It takes text, image and video input and returns text.

Amazon's page

GPT-4o mini (OpenAI)

Thu, 18 Jul 2024 00:00:00 GMT

OpenAI released GPT-4o mini on July 18, 2024.

GPT-4o mini is a smaller, lower-priced model from OpenAI, released on July 18, 2024 (snapshot gpt-4o-mini-2024-07-18). It takes text and image input and returns text.

OpenAI's page

GPT-4o (OpenAI)

Mon, 13 May 2024 00:00:00 GMT

OpenAI released GPT-4o on May 13, 2024.

GPT-4o is a multimodal model from OpenAI, first released on May 13, 2024 (snapshot gpt-4o-2024-05-13, which OpenAI shuts down in the API on October 23, 2026). In the API it takes text and image input and returns text.

OpenAI's page

Command R+ (Cohere)

Thu, 04 Apr 2024 00:00:00 GMT

Cohere released Command R+ on April 4, 2024.

Command R+ is a 104B-parameter language model from Cohere, announced on April 4, 2024 for enterprise retrieval-augmented generation (RAG) workloads. Its weights are published for non-commercial use.

Cohere's page