--- title: "Gemma 4 in Claude Code — local, or for a whole team · Selan" description: "Gemma 4 runs in Claude Code through Ollama, locally or on its cloud. Run it for a team on Ollama Cloud or OpenRouter, beside Claude, no key on a laptop." url: https://selan.ai/guides/gemma-in-claude-code updated: 2026-11-10T09:52:21+02:00 --- Guide · Other models in Claude Code # How do you run Gemma 4 in Claude Code? Run `ollama launch claude --model gemma4`. Ollama starts Claude Code on Gemma 4 through its Anthropic-compatible API, on your own machine or on Ollama's cloud. Through Selan, an owner connects the company's Ollama Cloud plan or OpenRouter key once. Each developer then runs `selan claude --model ollama:gemma4:31b`. ## Why does Gemma 4 work in Claude Code at all? Claude Code sends every request in Anthropic's Messages format. It reaches another server when `ANTHROPIC_BASE_URL` is set to that server's address. Ollama answers in that format, on your own machine and on its cloud. OpenRouter answers in it too. Claude Code edits files and runs commands through tool calls. Google built Gemma 4 with native function calling, so it can drive those tools. Google publishes Gemma 4 as open weights under the Apache 2.0 license. Anyone may run it, on their own hardware or on a host's. ## How do you run Gemma 4 in Claude Code on your own machine? Install Ollama and Claude Code, then start Claude Code on Gemma 4: ```bash ollama launch claude --model gemma4 ``` A bare `gemma4` is Ollama's default tag, the small E4B model with a 128K window. For a larger repository, Ollama says to set the context length to 64K or higher. The model then runs on your own hardware, and Ollama's plans put no limit on local use. How fast it answers depends on that hardware. For one developer with a capable machine, this is the simpler answer. A gateway adds nothing there. ## Which Gemma 4 size should Claude Code use? Gemma 4 comes in five sizes. Ollama names each one as a tag: | Ollama tag | What it is | Window | | --- | --- | --- | | `gemma4:e2b`, `gemma4:e4b` | Small models made for laptops and phones | 128K | | `gemma4:12b` | A medium model | 256K | | `gemma4:26b` | A mixture-of-experts model with 4B parameters active | 256K | | `gemma4:31b` | The largest, a dense model | 256K | | `gemma4:cloud` | Ollama's cloud, with no download | 256K | The 26B model runs almost as fast as a 4B model, Google says, because only 4B of its parameters work on each token. The 31B model is a download of about 20GB. Ollama's cloud runs it without one. ## How do you use Gemma 4 on Ollama's cloud? With the Ollama app, name the cloud tag: ```bash ollama launch claude --model gemma4:cloud ``` Without the app, create an API key on ollama.com. Set `ANTHROPIC_BASE_URL` to `https://ollama.com` and put the key in `ANTHROPIC_AUTH_TOKEN`. Then name the model as Ollama's cloud lists it, `gemma4:31b`. Your code then leaves your machine for Ollama's servers. The usage counts against your Ollama plan. On a team, that is one Ollama key and one set of variables per laptop, with no shared view of who used what. Why a key on every laptop is the expensive part is in [keeping API keys off laptops](https://selan.ai/guides/claude-code-without-api-keys). ## How does a team run Gemma 4 in Claude Code through Selan? Gemma 4 is a model Selan lists, at Ollama as `gemma4:31b` and at OpenRouter as `google/gemma-4-31b-it`. An owner connects the account once, then switches Gemma 4 on under [Settings → Models](https://control.selan.ai/settings/models). Developers install the `selan` CLI, sign in as themselves and name the model: ```bash selan claude --model ollama:gemma4:31b ``` The part before the first colon is the provider. Everything after it is that provider's own model id, passed through as it is. With `--model`, Selan tells Claude Code the model's real context window. Selan lists Gemma 4 at 262,144 tokens, so a session compacts there and not at 200K. If an owner set a lower context limit for your company, it compacts at that instead. To skip the flag, an owner adds Gemma 4 to Claude Code's picker under [Settings → Model pickers](https://control.selan.ai/settings/pickers). It then appears in `/model` beside the Claude models. The details are in [Run any model](https://selan.ai/docs#any-model). ## Which account does a Gemma 4 turn run on? | Route | How it bills | Name it as | | --- | --- | --- | | An Ollama Cloud plan | Your Ollama plan's monthly usage, then any credits you bought | `ollama:gemma4:31b` | | An OpenRouter key | Per token, on your OpenRouter invoice: $0.20 per million input tokens and $0.40 per million output tokens | `openrouter:google/gemma-4-31b-it` | A Gemma 4 turn runs on the provider its name points at, and on no other. If that account is out of quota, the request fails rather than moving to the other one. Selan reaches Ollama's cloud at ollama.com, never an Ollama server on a laptop. How Ollama itself runs in Claude Code is in [using Ollama with Claude Code](https://selan.ai/guides/ollama-in-claude-code). Every provider is on [Providers and models](https://selan.ai/docs/providers). ## What does a team get beyond the model? - **One usage record.** Gemma 4 turns land beside Claude turns, per person and per repository. [Usage per developer](https://selan.ai/guides/claude-code-usage-per-developer) shows what an owner sees. - **One spend limit.** An OpenRouter turn is metered, so it counts against a member's limit. A turn on an Ollama plan does not. See [spend limits](https://selan.ai/solutions/spend). - **No Ollama or OpenRouter key on a laptop.** Keys stay on the gateway. A developer holds a Selan session you can revoke. - **Codex on the same plan.** `selan codex --model ollama:gemma4:31b` uses Ollama's own Responses API. See [Codex](https://selan.ai/docs/codex). - **Cheap agent work.** Selan's comment-cleaning agent names `openrouter:google/gemma-4-31b-it` as its model, because the work is mechanical. See [the minimise-comments agent](https://selan.ai/docs/agents/minimise-comments). ## What should you check before switching a team to Gemma 4? - **Local turns are not on the record.** A developer running Gemma 4 on a laptop by Ollama's own setup bypasses Selan. Its usage record and spend limits do not see those turns. - **The size is part of the answer.** Selan runs the 31B model. A teammate comparing it with E4B on a laptop is comparing two different models. - **The model must be switched on.** A model your company cannot use is refused before Claude Code starts, not at the first request. **Good to know:** Anthropic doesn't support routing Claude Code to non-Claude models through any gateway, Selan included. A Gemma 4 session is the gateway's to answer for, not Anthropic's. [Start for free](https://auth.selan.ai/r) Free for up to three seats, and the Free plan needs no card. ### The short version - Ollama answers in Anthropic's API format, so `ollama launch claude --model gemma4` runs Claude Code on Gemma 4. - Gemma 4 is open weights under Apache 2.0, in five sizes; the small ones have a 128K window and the rest 256K. - On a laptop, Ollama puts no limit on it; for one developer with the hardware, that is the simpler answer. - Through Selan, a developer runs `selan claude --model ollama:gemma4:31b`, or `openrouter:google/gemma-4-31b-it` per token. - Gemma 4 runs on your Ollama plan or OpenRouter key, with one usage record and no key on a laptop. The sizes, windows, license and function calling are Google's Gemma 4 model card. The tags, the launch command and the context advice are Ollama's Gemma 4 page and its Claude Code and cloud docs. The OpenRouter price and window are its model list. All were read on 10 November 2026, as was Ollama's cloud model list. Anthropic's position is its gateway docs, as read on 3 November 2026. What Selan does is its model list, its CLI and its providers, Codex, agents and getting-started docs as they stood on 10 November 2026.