What's New
We list notable user-facing improvements here. For model/version details, see Model Matrix.
February 9, 2026 — New Claude 4.6 Opus, Grok Code Fast, and Kimi K2 Models
Added Anthropic Claude 4.6 Opus - latest flagship model with 200K context (1M with beta header), top-tier reasoning and coding performance.
Added xAI Grok Code Fast - economical reasoning model optimized for agentic coding with 256K context.
Added Moonshot Kimi K2 (via Groq) - trillion-parameter MoE model with 32B active parameters, 256K context, excellent for agentic tasks.
Marked Claude 4.5 Opus as legacy.
December 15, 2025 — xAI Grok 4.1 Fast Models
Added xAI Grok 4.1 Fast (Reasoning) - optimized tool-calling model with a massive 2M token context window and advanced reasoning capabilities for complex agentic tasks. Features 50% reduction in hallucinations and Agent Tools API support.
Added xAI Grok 4.1 Fast (Non-Reasoning) - ultra-fast variant with 2M context window for immediate responses without thinking tokens, perfect for high-throughput applications.
Note: The standard Grok 4.1 model is available on grok.com and X apps but not via API. API users have access to the Grok 4.1 Fast variants.
Marked older Grok models (2, 3, 3-mini, and 4) as legacy.
December 14th 2025 — Speech Input with Voice Commands
Added microphone support for speech-to-text input across AI Assistant, Assignments, and Insights. You can now speak your queries and use voice commands like "submit", "send", or "go" to automatically send your message.
December 13th 2025 — New GPT-5.2 model and model filtering
Added OpenAI GPT-5.2 model support
Better model selection for filtering out older models
December 1 2025 — New Anthropic Claude Opus 4.5
Added Anthropic Claude Opus 4.5 model support
November 30 2025 — Add Settings for AI
The Settings menu lets you override the prompt, tone, verbosity, and other detailed AI parameters. For instance, you can add to the prompt "Refer to the user as Dave."
November 18 2025 — New Gemini model
Added Gemini 3 Pro Preview model support
November 16, 2025 — New and updated models and file save enhancement
Added OpenAI GPT-5.1 model support
Add Meta LLaMA 3.3 via Groq
Added OpenAI gpt-oss-120b open source via Groq
Put file save/load tools explicitly in tool list
October 19, 2025 — Claude Sonnet and Haiku 4.5 Models
Added Claude Sonnet 4.5 model support
Added Claude Haiku 4.5 model support
Sept 21, 2025 — X Twitter integration and Claude 4.1
X Twitter integration to search, post, like and many twitter operations. Requires X developer keys.
Added Claude 4.1 model support
Updated — DeepSeek models to V3.1
Improved reasoning, long-context handling, and tool/agent performance.
Aug 23, 2025 — AI Posts + DeepSeek V3.1
Added — Automatically generated AI Posts
Fresh, AI-generated articles you can browse by topic—great for inspiration and to showcase what the AI can do.
Updated — DeepSeek models to V3.1
Improved reasoning, long-context handling, and tool/agent performance.
Aug 8, 2025 — OpenAI GPT-5 models
Added — GPT-5 across the app
We enabled GPT-5 (and GPT-5 mini, where applicable) the day after OpenAI’s public release (Aug 7, 2025).
Expect stronger coding, reasoning, and tool use, with adjustable “thinking” depth.
Aug 1, 2025 — Enhanced documentation
Added — Overhauled online docs
New Getting Started, Product Tour, and guides for Tasks, Files, and Integrations.
Jul 1, 2025 — MCP tools
Added — Model Context Protocol (MCP) support
Bring your own tools/data into Fresh Focus AI via MCP connectors for richer, personalized workflows.
Earlier highlights
- Ongoing UX polish, performance upgrades, and reliability improvements.
- Expanded model roster across OpenAI, Anthropic, Google, DeepSeek, Cohere, xAI, and more.
Thanks for using Fresh Focus AI!