AIHubMix Desktop 1.0: one setup layer for 11 coding agents, plus DeepSeek V4.1 Flash 1017 at 50% off

Pointing a single coding agent at a gateway is a ten-minute edit. Pointing four of them at the same gateway — and keeping them there across two version bumps — is maintenance nobody schedules. When it goes wrong, it goes wrong quietly: the agent keeps answering, just from a provider you did not intend, billed to an account you did not mean to use. Or the key ends up pasted into a file that later gets committed.

AIHubMix Desktop 1.0.0 is out for macOS 12+ and Windows to close that gap. It scans the machine for coding agents you already have installed, writes the configuration each one needs, keeps the credential in the OS keychain, and on disconnect removes only what it wrote. It is not a chat client — there is no conversation window. It is the setup layer for agents you already run, and one key covers more than 700 models.

What the app does to a machine

Scan. On launch the app looks for supported agents, reports the version and install path of each, and marks whether the installation is usable. What it offers is what it finds: an agent that is not installed does not appear in the list.

Connect. It writes the endpoint, the model list and the credentials an agent needs, and leaves the rest of the file alone.

Verify. Connecting sends a small billable test request for each model it sets up, and checks that the request really arrived at the gateway. An HTTP 200 from the wrong provider on the other end does not count as verified.

Switch. Model, default model, reasoning effort and approval mode are set from the app rather than from a file. Reasoning levels are filtered per model — only levels the model declares are offered, with the full catalog on the models page. Codex CLI and Codex Desktop share one config file, so a change made for one applies to both.

Disconnect. Only the app's own writes are removed. Your edits to those files stay, the previous settings come back, and there is no restore-from-memory step because the app recorded what it replaced.

Two connection styles run underneath. Some agents take a whole provider entry, which puts the model list and default in one place; others take a single endpoint and key, and the model is chosen inside the tool. The app keeps the two apart per agent, so connecting one never rewrites the settings of another. Where two tools share a file, the change is shown once and applied to both.

Keys go into the macOS Keychain or Windows Credential Manager, not into a plaintext file in your home directory. That is the part most manual setups get wrong, because the path of least resistance is export ANTHROPIC_AUTH_TOKEN=... in a shell profile.

Prerequisites

• macOS 12 or later, or Windows. Both Mac builds are signed and notarized. Windows installers are not Authenticode-signed yet, so SmartScreen may ask you to confirm the publisher, and the WebView2 runtime may need an internet connection during setup.
• An AIHubMix account. The app holds a client key for the agents it configures; if that key is ever revoked or leaked, it can be reissued and rewritten into the connected agents from the app.
• The agents you want to connect, already installed. The app cannot configure what is not on disk, and Codex CLI plus the CLI inside Codex Desktop must be close enough in version to share a model list.

Connect an agent

1. Download the installer for your platform from the latest release.
2. Sign in with your AIHubMix account.
3. Choose the agents to connect and press Connect. The app writes the fields the gateway needs — endpoint, model list, credentials — and leaves the rest of each file as it is.
4. Restart whatever caches its configuration. The next section is about that step, because "restart the app" is not always enough.

At launch the app detects and configures eleven agents, among them Claude Code, Claude Desktop, Codex (the CLI and the desktop app are managed together), Gemini CLI, OpenCode, MiniMax Code, OpenClaw, Hermes and Pi.

Manual setup still works, and the app is a shortcut over exactly that configuration: Claude Code, Codex and Gemini CLI each have a docs page with the same edits written out.

The commands that still bite

Most tools read their configuration at startup, so a successful Connect is not a successful switch until the reader has restarted. Two cases need more than clicking around:

# Codex's background service holds its own copy; restart it before new terminal sessions
codex app-server daemon restart

Claude Desktop has to be fully quit, not just closed: closing the window is not enough for it to pick up the new configuration.

The other trap is a command you did not write in this session at all:

# In ~/.zshrc or ~/.bashrc — takes priority over the config file the app writes
export ANTHROPIC_AUTH_TOKEN=...

A base URL or key exported by the shell beats the file on disk, and the terminal wins. The app flags the conflict rather than silently losing the fight; the fix is to remove the export from your profile.

Failure modes and fixes

What the connection check does and does not prove

The verification is a test request that confirms the request reached the gateway — provider identity, not performance. Nothing in it reports latency, throughput or behaviour under load, and this post gives no measured figures for any of the three. Read a green check as "the wiring is right", not as "the agent is fast".

Two things also stay outside the app's control: what a running agent does with a configuration it has already loaded, and whether a locked-down machine will let the file be written at all.

The launch promotion: DeepSeek V4.1 Flash 1017 at 50% off

To mark the launch, DeepSeek V4.1 Flash 1017 is 50% off for two weeks, through 12 November 2026. Both price segments are halved, cached reads included.

Peak hours are 01:00–04:00 and 06:00–10:00 UTC; the rest of the day is off-peak. It is the same model on the same endpoint, deepseek-v4.1-flash-1017 — a 1M-token context, up to 393K output tokens, vision input, tool calling and prompt caching.

For context, DeepSeek's own list price is $0.15 per million input and $0.60 per million output at off-peak, doubling at peak. During the promotion the gateway's input rate is about half of DeepSeek's own off-peak rate.

The promotion is a reasonable place to point an agent that reads a lot, so the worked example is a cache-heavy one. One heavy day of coding-agent work: 20M input tokens with 18M of them served from cache, plus 2M output tokens.

That is the per-token prices above multiplied by the token counts in this paragraph; your own split will differ. The promotional rate is applied automatically to requests for that model ID until 12 November 2026 — no code, no coupon — and billing returns to list prices afterwards.

Get the app

Download AIHubMix Desktop, connect the agents you already run, and point the one that reads the most at DeepSeek V4.1 Flash 1017 while the half-price window is open. The app is free; your agents' requests bill to your AIHubMix balance at normal per-token prices, the same as any API call, and the balance and today's spend are visible inside the app. The half-price window closes on 12 November 2026.

Sources

• AIHubMix Desktop 1.0.0 release notes — version, platforms, signing, updater
• AIHubMix desktop download — installers, supported platforms, model count
• DeepSeek V4.1 Flash 1017 on AIHubMix — specifications and pricing, including the promotional offers and their validity date
• AIHubMix docs: Claude Code, Codex, Gemini CLI — manual setup for the agents the app also configures
• DeepSeek API change log — the vendor's own model naming and list prices