NextCoWork · Desktop AI agent

Bring AI
into your workflow.

Read, search and execute in your local workspace.Connect to multiple models. Pay for the calls you make.

Free desktop app · Usage-based model billingWindows · x64 · Picked for your device

WindowsmacOSLinuxAll platforms
ImageNextCoWork
Interface demo
Project analysis

Review this project and suggest improvements.

Agent
  1. Read project structure

    Map files and workspace context

    Completed
  2. Search key modules

    Understand code and dependencies

    Completed
  3. Outline improvements

    Turn the findings into actionable next steps

    In progress
Workspace analysis
  1. Map the modules and their dependency boundaries.
  2. Add tests around critical logic to verify changes.
  3. Outline next steps for you to review.

Interface demo only. No files are read or model requests sent.

Built for real work

One workspace. A connected toolchain.

From understanding a project to acting on it,
keep files, tools and models in the same context.

Local workspace

Files, code and context

Agent tool loop

From understanding to execution

Unified model gateway

Mainstream API compatibility

Your local folder is the workspace

Open a directory and the agent reads, searches and executes inside it — path fencing keeps it from wandering out.

  • Open any local folder
  • Path fencing bounds every read and write
  • Remote SSH workspaces supported
DOWNLOAD

Cross-platform. Start in your workspace.

One workflow across Windows, macOS and Linux.
Find installers, release information and SHA-256 checksums for every platform.

v2.8.4Published Oct 18, 2026SHA-256 51a16263e7a0…

See every platform and checksum
MODEL CATALOG

Every model the app can pick,
priced on the same card.

Context, max output, tool calling and vision,
See the model catalog for current availability and effective prices.

The cards below are interface examples, not live models, prices or operational status.

OpenAI

GPT-6 Sol

gpt-6-sol
Chat
74% off
InputUSD 4USD 1.04per million tokens
OutputUSD 20USD 5.2per million tokens
Cache hitUSD 0.4USD 0.104per million tokens
Context400KMax output128K
Loading…
ReasoningVision input+2
OpenAI

GPT-6 Luna

gpt-6-luna
Chat
74% off
InputUSD 0.2USD 0.052per million tokens
OutputUSD 1.2USD 0.312per million tokens
Cache hitUSD 0.02USD 0.0052per million tokens
Context262KMax output64K
Loading…
ReasoningVision input+1
See every model
DROP-IN COMPATIBLE

No desktop app?
Change one base_url instead.

The same models are served over three mainstream protocols.
Whichever one your client already speaks, use that.

BASE URLhttps://nextco.work
/v1/messages

Works directly with the official Claude SDK and Claude Code

curl https://nextco.work/v1/messages \
  -H "x-api-key: $COWORK_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{"model":"...","max_tokens":1024,"messages":[{"role":"user","content":"hello"}]}'
TRANSPARENT BY DEFAULT

Every dollar,
traceable.

Accounts are free; models bill by usage.
Usage and balance both live in the console.

Priced per line item

Input, output and cache hits are priced separately; some models add tiers and time-of-day rates, and the bill settles at whichever tier actually applied.

USD wallet

Cash balance and granted credit are tracked separately, and holds in flight get their own line instead of blurring into one number.

API key quotas

Each key can carry its own monthly and total caps plus an expiry date. Past the cap it is refused — never silently overdrawn.

GOOD QUESTIONS

Before you start,
you might be wondering.

Install the app first,
decide about the API later.