Sparkian launches a $19 AI workspace: 100+ models in one chat

Geekflare’s Sparkian has launched a $19 monthly Pro plan that puts more than 100 AI models — from OpenAI, Anthropic, Google and others — inside a single chat window. The announcement, issued from London on October 17, 2026, is aimed squarely at subscription fatigue: Sparkian compares its price with the roughly $60 a month it costs to hold separate ChatGPT, Claude and Gemini subscriptions at about $20 each.

The draw is not just the price. In Sparkian, the conversation stays put while the model changes — research with Claude, draft with GPT, generate the image with something else, all in one thread. The company says the aim is to end the familiar ritual of managing several subscriptions and rebuilding context every time you switch tools.

Many models, one conversation

Sparkian’s chat supports models from OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, Grok, Qwen, Kimi and Perplexity. Its signature feature is side-by-side answers: send one prompt to GPT, Claude and Gemini at once, compare the replies, and continue the conversation with whichever answer won. Switch models mid-thread and your messages, files and earlier decisions stay in place.

The workspace also pulls in web pages or files from your knowledge base and attaches citations, runs web search automatically when a question needs it, and lets teams work together on AI tasks in Projects or Workspaces.

A credit system instead of unlimited everything

The Pro plan runs on credits, not unlimited use. You get 10,000 Sparks a month, spent from a single balance across both chat and image generation. Sparkian’s own usage estimates put light models like Qwen 3.8 Flash at about 1 Spark per request and top-tier ones like Claude Opus 5 at 25 Sparks — so the monthly allowance covers up to 10,000 light requests or roughly 400 heavy ones.

The image studio lives in the same chat or in a dedicated studio, with image models including Nano Banana, GPT Image, Seedream, Recraft, Grok Imagine and Flux. Size, quality and count are adjustable, and a reference option lets you attach a generated image and ask for a change — a different color, pose or detail — to produce a new version.

Need more than the monthly allowance? One-time Spark packs run from $15 for 5,000 Sparks to $200 for 100,000 and stay valid for a year. Extra seats cost $5 a month each, and there is a free plan with 100 Sparks that requires no card. Geekflare’s own hands-on guide adds that Pro includes two seats, two workspaces, a 100 MB knowledge base, unlimited in-chat web search and a year of chat history.

The fine print is the product

Sparkian’s announcement is refreshingly direct about the trade: credits are not unlimited. Advanced models, long conversations and attachments burn Sparks faster than the starting estimates, and image costs vary by model — though an estimate is shown before each request. Users who mainly compare answers or generate the occasional image will likely spend less than on three separate subscriptions; heavy users of top-tier models should watch the burn rate.

“People shouldn’t have to manage several subscriptions and rebuild context every time they switch tools,” said Chandan Kumar, Sparkian’s founder. “With Sparkian, the conversation stays in place while the model changes.”

Whether one $19 subscription survives the habits of people who already pay for ChatGPT Plus is an open question — but Sparkian’s bet is that the comparison view, not the discount, is what keeps them.

• Who: Sparkian, an AI workspace built by London-based Geekflare; founder Chandan Kumar.
• What: a $19/month Pro plan putting 100+ models from OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, Grok, Qwen, Kimi and Perplexity in one chat app.
• When: announced October 17, 2026.
• Credits: 10,000 Sparks a month for chat and image generation from one balance — Qwen 3.8 Flash costs about 1 Spark per request, Claude Opus 5 about 25, so the allowance covers up to 10,000 light requests or roughly 400 top-tier ones.
• More pricing: one-time Spark packs from $15 (5,000 Sparks) to $200 (100,000 Sparks), valid a year; extra seats $5/month each; free plan with 100 Sparks, no card required.
• Also in Pro (per Geekflare's guide): two seats, two workspaces, a 100 MB knowledge base, unlimited in-chat web search and a year of chat history.

DeepSeek pushed its newest checkpoint, V4.1-Flash-1017, live across its web chat, app, and API on October 17, 2026 — a "minor update" to V4.1 Flash that the company says cuts task execution time and output length while lifting scores across reasoning, coding, agentic, and visual benchmarks. The weights are MIT-licensed, and DeepSeek used the release to open its in-house Search and Fetch APIs to everyone for the first time.

The headline number: 77.0 on DeepSWE

The model card's benchmark table puts DeepSWE v1.1 at 77.0 — up from 74.2 on the previous Flash checkpoint, and ahead of Claude Opus 5 (74.0) and GPT-5.6 Sol (73.0) on that specific coding benchmark. Terminal-Bench 4.0 jumps to 42.3 from 31.2, a sharp move for agent-terminal work. MarkTechPost's read of the release material reports the new checkpoint beats its predecessor on all 11 benchmarks DeepSeek published, by 2.0 to 11.1 points.

Elsewhere on the card: HLE without tools at 42.6, NL2Repo-Bench at 68.2, Apex Bench at 72.1, BabyVision with tools at 92.5, and Automation-Bench at 60.5. DeepSeek ran Terminal-Bench on the Claude Code harness and DeepSWE on Mini-SWE-V2.

Coding yes, raw reasoning no

The fine print matters. On pure reasoning, the frontier still leads: HLE without tools comes in at 42.6 for Flash-1017 versus 56.3 for Opus 5, and GPT-5.6 Sol edges it too at 44.5. This release wins the agent-coding lane — autonomous software work, terminal tasks, repo-level generation — not the raw-reasoning crown. DeepSeek frames it the same way, calling the update an advance of the "performance–cost Pareto frontier across multiple agent harnesses."

Price unchanged — that's the point

The API name deepseek-flash and its pricing are unchanged: $0.15 per million input tokens and $0.60 per million output tokens, with cache reads at $0.003 per million. The context window stays at 1,048,576 tokens with up to 393,216 completion tokens, and the model card recommends temperature 1.0 and top_p 1.0. Faster tasks and shorter outputs on the same price card is the whole pitch: better Pareto, not a new price.

A live Search API, priced to be used

Alongside the weights, DeepSeek opened its Search and Fetch APIs — backed by its in-house search engine — to everyone at $0.75 per 1,000 successful requests, capped at 5 QPS per account. Server-side calls routed through DeepSeek Harness, Claude Code, or Codex stay free through the end of October 2026, after which the rate card applies. It's the company's first public rate card for live web search inside its own models — a direct play at the agent-search lane.

Under the hood: KV cache compression

Technical coverage of the release describes the checkpoint as the first built on DeepSeek's Causal Encoder-Decoder (CED) architecture — a 552B-parameter sparse mixture-of-experts design with asymmetric activation to keep per-token compute low. The companion paper, titled "DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression," puts the memory story front and center: compressing the KV cache is what lets a model this size serve agent workloads at fifteen cents a million tokens. The company also says reinforcement learning with self-feedback improved end-to-end project building, such as game creation and 3D modeling.

Why it matters

Agent coding is where inference spend actually lands — long multi-step trajectories, terminal loops, tool calls — and DeepSeek just made its strongest entry in that lane cheaper per task without touching the price card. MIT weights mean the 552B checkpoint can be served, distilled, and studied anywhere the hardware allows, and the community moved fast: a GGUF conversion sized for a single 128GB DGX Spark appeared on Hugging Face the same day. The open question is how quickly rivals answer the DeepSWE number — and whether anyone answers the Search API.

Sources: the official DeepSeek-V4.1-Flash-1017 model card on Hugging Face (benchmarks, license, sampling parameters), AI Weekly's release coverage (Search API rate card, MarkTechPost benchmark deltas), and OpenRouter's model page (pricing, context window, API name).