> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.sparkian.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.sparkian.com/_mcp/server.

# Changelog

New features, model updates, and fixes as they ship.

## October 19, 2026

## Image studio

Sparkian now has a dedicated place to create images. Open **Images** from the sidebar to generate with models from providers like Nano Banana, GPT, Seedream, and Recraft, pick an aspect ratio and style, generate up to four images at once, and turn on **Compare models** to see how different models handle the same prompt.

Your images are saved in a new **Images** tab in your Library. You can regenerate any image, use it as a reference for the next one, or save it to a project.

See [Image studio](/image-studio) for how it works. You can still generate images inside a chat, see [Creating images](/create-image).

## New model: Gemini 4 Argon

Gemini 4 Argon is now available in the model picker. See [Sparks (credit system)](/sparks) for how each model's cost is estimated.

## September 29, 2026

## Listen to a response

Every response now has a **Listen to response** action, so you can have it read aloud instead of reading it yourself. See [Chat response actions](/chat-response-actions).

## Jump between prompts in a conversation

Conversations with more than one prompt now show a navigator along the edge of the chat — hover a mark to preview that prompt, or click to jump straight to it. See [Jumping between prompts](/conversation-prompts).

## September 28, 2026

## Context modes: Smart and Exact

Chat settings now include a **Context mode** option alongside Recent messages. **Smart** (default) summarizes older messages, which can use fewer Sparks in long chats. **Exact** sends only your selected recent messages, exactly as written, with no summarizing. Change it anytime for the current conversation in **Chat settings**.

See [Memory and context](/memory-and-context) for how it works.

## Personal memory

Sparkian can now remember useful details about you — your role, preferences, how you like answers formatted — across your private chats in a workspace, so you don't have to repeat yourself in every new conversation. Turn it on or off, and manage what's saved, from **Settings → General → Memory**.

See [Personal memory](/personal-memory) for details.

## September 24, 2026

## New models

Added the following models to the model picker:

* Claude Opus 5.5
* GPT-6 Sol
* GPT-6 Luna
* DeepSeek V4.1 Flash

See [Sparks (credit system)](/sparks) for how each model's cost is estimated.

## Workspace analytics

**Settings → Analytics** now shows how your workspace uses Sparkian: messages and Sparks over time, your most active chats and members, which models are used most, and what your workspace contains. Filter by date range, model, member, or project. See [Analytics](/analytics).

## Sparks consumed per message

Every response now shows how many Sparks it used. Hover the ✦ icon in a response's action row to see the exact amount. See [Sparks (credit system)](/sparks).

## September 18, 2026

## Use a generated image as a reference

You can now take any image Sparkian generates and use it as a reference for your next generation — describe the change you want (a different color, pose, or detail) and Sparkian generates a new image based on it, instead of starting from scratch.

See [Creating images](/create-image) for how it works.

## Python for data analysis and charts

Sparkian now detects when a question needs real calculation — analyzing a spreadsheet, computing statistics, plotting a trend — and runs Python behind the scenes to answer it, showing the code it ran alongside the result and any charts it generates. Works with CSV, Excel, or data pasted directly into a message.

See [Data analysis and charts](/data-analysis) for how it works.

## September 13, 2026

## Delete your account

You can now permanently delete your Sparkian account, under **Settings → General → Danger zone**. See [Account settings](/account-settings) for what gets deleted and how workspace ownership is handled.

## September 11, 2026

## New models

Added the following models to the model picker:

* Fable 5.1
* GPT-6 Astra
* Gemini 3.8 Flash
* Qwen 3.8 Flash

See [Sparks (credit system)](/sparks) for how each model's cost is estimated.

## September 9, 2026

## Geekflare Chat is now Sparkian

Geekflare Chat has migrated to **Sparkian** — a new name for the same AI workspace, with more on the way. Existing accounts, chats, and workspaces carry over as-is.

See [Welcome to Sparkian](/welcome) for an overview of what's here today.