Every AI release note

DeepSeek · DeepSeek API·Jul 25, 2024

New API Features

Update API /chat/completions JSON Mode. Function Calling. Chat Prefix Completion（Beta）

Google · Gemini app·Jul 25, 2024

Introducing Gemini 1.5 Flash to Gemini, plus 2 more changes

What: Gemini now gives you access to 1.5 Flash, our model optimized for speed and efficiency. 1.5 Flash provides a 4x longer context window (from 8k tokens to 32k tokens) compared to 1.0 Pro and improved overall response quality and accuracy. Why: 1.5 Flash provides a longer context window and improved performance to unlock new ways to create, interact and collaborate with Gemini. A longer context window allows Gemini to process a larger amount of content for each prompt, helping you solve more complex problems and boosting your productivity when you explore and analyze information.

DeepSeek · New model·Jul 24, 2024

deepseek-coder

The deepseek-coder model has been upgraded to DeepSeek-Coder-V2-0724.

Mistral AI · Mistral API·Jul 24, 2024

We added fine-tuning support for Codestral, Mistral Nemo and Mistral Large

Now the model choices for fine-tuning are open-mistral-7b (v0.3), mistral-small-latest (mistral-small-2402), codestral-latest (codestral-2405), open-mistral-nemo and, mistral-large-latest (mistral-large-2407)

Mistral AI · New model·Jul 24, 2024

We released Mistral Large 2

We released Mistral Large 2 (mistral-large-2407).

Mistral AI · New model·Jul 18, 2024

We released Mistral Nemo

We released Mistral Nemo (open-mistral-nemo).

Mistral AI · New model·Jul 16, 2024

We released Codestral Mamba and Mathstral

We released Codestral Mamba (open-codestral-mamba) and Mathstral.

Anthropic · Claude API·Jul 15, 2024

Generate outputs up to 8,192 tokens in length from Claude Sonnet 3.5 with the new anthropic-beta:…

Google · Gemini app·Jul 15, 2024

Workspace Extensions now available in Beta for Gemini for Google Workspace add-on subscribers

What: Google Workspace admins can now opt-in to beta test Gemini Extensions for Gmail and Drive for their organization. Via the Drive extension, organizations can also access information from Google Docs. This enables users to find and summarize information, as well as ground Gemini’s responses, from their content. Google Workspace data will not be used to train Gemini’s public model, and admins can disable access at any time. Subscription to a Gemini for Google Workspace add-on is required. Why: Now, Google Workspace users can collaborate with information from select Google Workspace apps, all in one place, bringing Gemini’s capabilities into user’s daily workflows, enhancing productivity and collaboration.

Google · Gemini API·Jul 12, 2024

Support for Gemini 1.0 Pro Vision removed from Google AI services and tools

Support for Gemini 1.0 Pro Vision removed from Google AI services and tools.

Anthropic · Claude API·Jul 9, 2024

Automatically generate test cases for your prompts using Claude in the Developer Console

Compare the outputs from different prompts side by side in the new output comparison mode in the Developer Console.

DeepSeek · New model·Jun 28, 2024

deepseek-chat

The deepseek-chat model has been upgraded to DeepSeek-V2-0628. Model's reasoning capabilities have improved, as shown in relevant benchmarks:

Anthropic · Claude API·Jun 27, 2024

View API usage and billing broken down by dollar amount, token count, and API keys in the new Usage and Cost…

View your current API rate limits in the new Rate Limits tab in the Developer Console.

Google · Gemini API·Jun 27, 2024

General availability release for Gemini 1.5 Pro's 2M context window

General availability release for Gemini 1.5 Pro's 2M context window. Added support for code execution.

Anthropic · Claude API·Jun 20, 2024

Claude Sonnet 3.5, our most intelligent model yet, is now available across the Claude API, Amazon Bedrock,…

Google · Gemini API·Jun 18, 2024

Added support for context caching

Added support for context caching.

Google · Gemini app·Jun 18, 2024

Gemini Advanced features available in India, plus 2 more changes

What: We’re making more Gemini Advanced features available in India, including Google’s next-generation model, 1.5 Pro, together with exclusive capabilities such as the ability to upload your documents, analyze your data, and process significantly more information with a 1 million token context window. Why: As we continue to build Gemini responsibly, we’re expanding access to more countries and languages.

DeepSeek · New model·Jun 14, 2024

deepseek-coder

The deepseek-coder model has been upgraded to DeepSeek-Coder-V2-0614, significantly enhancing its coding capabilities. It has reached the level of GPT-4-Turbo-0409 in code generation, code understanding, code debugging, and code completion. Additionally, it possesses excellent mathematical and reasoning abilities, and its general capabilities are on par with DeepSeek-V2-0517.

Google · Gemini API·Jun 12, 2024

Gemini 1.0 Pro Vision deprecated

Gemini 1.0 Pro Vision deprecated.

Google · Gemini app·Jun 5, 2024

The Gemini mobile app is available in Europe

What: The Gemini app is now available in countries like the UK, Germany, France, Italy, Sweden, Switzerland, and more. You can download the Gemini app from the Android Play Store, or try it in the Google App on iOS. Chat to get help with writing, planning, learning and more in a variety of supported languages. More languages and countries will be coming soon. Learn more about language, country, and device availability in the Help Center. Why: We’re dedicated to bringing Gemini responsibly to more countries and languages, on the devices you use the most.

Mistral AI · Mistral API·Jun 5, 2024

We released fine-tuning API

Check out the capability docs and guides.

Anthropic · Claude API·May 30, 2024

Tool use is out of beta across the Claude API, Amazon Bedrock, and Vertex AI, with no beta header required

Mistral AI · New model·May 29, 2024

New model available: codestral-latest

Check out the code generation docs.

Google · Gemini API·May 23, 2024

Gemini 1.5 Pro is generally available

Gemini 1.5 Pro (gemini-1.5-pro-001) is generally available (GA). Gemini 1.5 Flash (gemini-1.5-flash-001) is generally available (GA).

Google · Gemini app·May 23, 2024

Enterprise-grade data protections for Workspace for Education customers is now available in Gemini

What: Available today, the new Gemini Education and Gemini Education Premium add-ons give Google Workspace for Education customers access to 1.5 Pro - Google’s most capable AI model widely available today, with enterprise-grade data protections. Administrators can manage Gemini settings within the Google Workspace Admin Console. The Gemini Education add-ons are available and optimized in English in over 150 countries and territories Why: With Gemini Education, education leaders, staff and students 18 years and older can save time, make learning more personal, inspire creativity and learn confidently in a private and secure environment. Gemini Education add-ons are covered under the Google Workspace for Education Terms of Service; your data is not reviewed by anyone, used to train AI models, or shared with other users or organizations.

Mistral AI · Mistral API·May 23, 2024

Function calling: tool_call_id is now mandatory in chat messages with the tool role

Function calling: tool_call_id is now mandatory in chat messages with the tool role.

Google · Gemini app·May 21, 2024

Exclusive to Gemini Advanced: Data Analysis

What: You can now upload your spreadsheets for faster data processing & exploration, presentation-ready charts, and deeper insights. Upload Google Sheets, CSVs, and Excel files directly from Google Drive or your device. Why: Efficiently process, explore, analyze, and visualize your data to save time and uncover insights faster.

DeepSeek · New model·May 17, 2024

deepseek-chat

The deepseek-chat model has been upgraded to DeepSeek-V2-0517. The model has seen a significant improvement in following instructions, with the IFEval Benchmark Prompt-Level accuracy jumping from 63.9% to 77.6%. Additionally, on API end, we have optimized model ability to follow instruction filled in the system" part. This optimization has significantly elevated the user experience across a variety of tasks, including immersive translation, Retrieval-Augmented Generation (RAG), and more. The model's accuracy in outputting JSON format has been enhanced. In our internal test set, the JSON parsing rate increased from 78% to 85%. By introducing appropriate regular expressions, the JSON parsing rate was further improved to 97%.