Groq

Groq Launches Responses API with OpenAI Compatibility

Groq has released a new Responses API in beta that is fully compatible with OpenAI's API, supporting text and image inputs, stateful conversations, and function calling. This enables easy integration of advanced conversational AI capabilities into existing applications. No immediate action required as this is a new optional feature in beta testing.

Date not specified

Groq releases GPT-OSS 20B and 120B open-source MoE models

Groq has released two new open-source Mixture-of-Experts language models (20B and 120B parameters) that compete with OpenAI's o4-mini and o3-mini models. These models feature reasoning capabilities, built-in browser search, code execution, structured outputs, and high performance metrics across various benchmarks. No action required - this is a new capability addition expanding Groq's model offerings.

Date not specified

Groq Python and TypeScript SDKs updated to v0.30.0 and v0.27.0

Groq has released new versions of their Python and TypeScript SDKs with unspecified improvements and changes. Developers using these SDKs should review the updates and consider upgrading to access the latest functionality. No breaking changes are indicated in this changelog entry.

Date not specified

Groq adds JSON schema support for guaranteed response formatting

Groq now supports structured outputs that guarantee model responses conform to provided JSON schemas, eliminating parsing errors and malformed responses. This new feature improves developer experience by removing the need for complex parsing logic and format prompt engineering. No action required - this is an optional enhancement for applications requiring strict data structure compliance.

Date not specified

Groq SDK updates improve OpenAI compatibility and fix message format errors

Python SDK v0.30.0 and TypeScript SDK v0.27.0 include improved chat completion message type definitions for better OpenAI compatibility. This update resolves errors that occurred with certain message formats. No action required, but updating is recommended for improved reliability.

Date not specified

Groq adds Kimi K2 Instruct - 1T parameter MoE model

Groq has released Kimi K2 Instruct, a new Mixture-of-Experts language model with 1 trillion parameters optimized for agentic intelligence, tool use, and coding tasks. The model features a 131K token context window and reportedly surpasses GPT-4.1 on agentic and coding benchmarks. This is a new capability addition that expands Groq's model offerings without requiring any immediate action from existing users.

Date not specified

Groq SDK Updates: Python v0.29.0 and TypeScript v0.26.0 Released

Groq has released updated versions of their Python and TypeScript SDKs with new features and improvements. Developers using these SDKs should review the changes and consider updating to access the latest functionality. This is a routine SDK update that enhances the developer experience.

Date not specified

Groq Python SDK updated to v0.28.0 and TypeScript SDK updated to v0.25.0

Groq has released new versions of their Python and TypeScript SDKs with updated functionality and code examples. This is a routine SDK update that may include bug fixes, performance improvements, or new features. Developers should review the changes and update their SDK versions when convenient.

Date not specified

Groq adds Qwen 3 32B model with advanced reasoning and multilingual support

Groq has released Qwen 3 32B, a new large language model featuring dual thinking/non-thinking modes, 128K context window, and support for 100+ languages. The model offers competitive pricing at $0.29/1M input tokens and $0.59/1M output tokens with strong performance across reasoning benchmarks. This is a new capability addition that expands available model options without requiring immediate action.

Date not specified

Groq updates Python and TypeScript SDKs to v0.26.0 and v0.23.0

Groq has released updated versions of their Python and TypeScript SDKs with new features and improvements. This is a standard SDK update that provides enhanced functionality for developers. No immediate action is required, but developers may want to upgrade to access the latest features.

Date not specified

Groq adds Llama Prompt Guard 2 models for prompt injection detection

Two new specialized classifier models (22M and 86M parameters) are now available to detect malicious prompt attacks like injections and jailbreaks. The models offer high accuracy rates (99.8% and 99.5% AUC) with significantly reduced latency compared to larger alternatives. This is an additive security feature that organizations can optionally integrate into their LLM applications.

Date not specified

Groq updates Python and TypeScript SDKs to v0.25.0 and v0.22.0

Groq has released new versions of their Python and TypeScript SDKs with updated functionality and code examples. This is a routine SDK update that likely includes bug fixes, improvements, and potentially new features. Developers using these SDKs should review the changes and consider upgrading to take advantage of the latest improvements.

Date not specified

Groq integrates Meta's Llama-Guard-4-12B content moderation model

Groq has released Meta's Llama-Guard-4-12B, a 12B parameter model specialized for content moderation across text and images with 128K token context. The model uses MLCommons Taxonomy framework to classify potentially harmful content in both user inputs and AI outputs. This is an additive feature that enhances content safety capabilities without requiring immediate action.

Date not specified

Groq adds domain exclusion and Wikipedia filtering to search

Groq has introduced new search settings that allow users to exclude specific domains from search results, with Wikipedia exclusion provided as a documented example. The feature includes advanced usage options with domain wildcards for more granular search control. This is an additive feature that enhances existing search capabilities without requiring any immediate action.

Date not specified

Groq updates Python SDK to v0.24.0 and TypeScript SDK to v0.21.0

Groq has released new versions of their Python and TypeScript SDKs with various improvements and updates. This is a routine SDK update that may include bug fixes, performance improvements, and new features. Developers using these SDKs should review the changes and consider upgrading to benefit from the latest enhancements.

Date not specified

Groq Python SDK updated to v0.23.0 and TypeScript SDK updated to v0.20.0

Groq has released new versions of their Python and TypeScript SDKs with various improvements and updates. Developers using these SDKs should review the changes and consider upgrading to access new functionality. This appears to be a standard SDK update with new features rather than breaking changes.

Date not specified

Groq enhances moonshotai/kimi-k2-instruct model with performance improvements and code examples

Groq has added performance enhancements to their platform and provided code examples demonstrating usage. This appears to be a feature addition that includes documentation and example implementations. No action required from users, but developers may benefit from reviewing the new performance-related code examples.

Date not specified

Groq adds Meta Llama 4 Scout and Maverick models with vision

Groq API now supports Meta's new Llama 4 Scout (17Bx16MoE) and Maverick (17Bx128E) models featuring multimodal capabilities including image understanding up to 5 images, 128K token context window, function calling, and JSON mode. This is a new feature addition that expands available model options without requiring any changes to existing implementations.

Date not specified