2B Gemma 4 Deployment with Cloud Run, NVIDIA L4, MCP SDK 2.x, and Claude Code

A developer published a step-by-step guide for deploying Google's 2B Gemma 4 model to Cloud Run with an NVIDIA L4 GPU, using vLLM for serving and a single-file Python MCP server to stage weights, deploy, health-check, benchmark, and tear down the service. The project also documents migrating the MCP server to the MCP Python SDK 2.x after an unbounded dependency caused a fresh pip install to break imports, as FastMCP was renamed to MCPServer.

This article provides a step by step deployment guide for Gemma 4 E2B to a Cloud Run hosted GPU enabled system. A suite of Python MCP tools is built to simplify management of the vLLM hosted deployment with Claude Code. https://github.com/xbill9/gemma4-dev/tree/main/gpu-2B-cloudrun-devops-agent https://github.com/xbill9/gemma4-dev/tree/main/gpu-2B-cloudrun-devops-agent This project is a DevOps/SRE assistant for a Gemma 4 model served by vLLM on Cloud Run with an NVIDIA L4 GPU. A single-file Python MCP server provides tools to stage the weights, deploy the service, check its health, benchmark it, and tear it down. Cloud Run is the serverless option. There is no VM to provision and no driver to install: the service scales to zero when idle, and one gcloud command attaches the GPU. Along the way the MCP server itself had to move to the MCP Python SDK 2.x, because a fresh pip install stopped it from starting. That migration is covered where it happened — in the MCP server section, before the deploy. The strategy for starting MCP development for model management is a incremental step by step approach. First, the basic development environment is setup with the required system variables and a working Claude Code configuration. Then, the Python MCP server is brought up over stdio and validated with Claude Code in the local environment. That server then stages the model, deploys it to Cloud Run, and drives validation and a benchmark sweep against the live endpoint. Requires-Python =3.10 us-east4 <project -bucket for the model weights Clone the repository and switch to the Cloud Run directory: cd ~ git clone https://github.com/xbill9/gemma4-dev cd gemma4-dev/gpu-2B-cloudrun-devops-agent Then run init.sh once. It checks your gcloud login and application default credentials, asks for a project ID, installs the Python requirements, enables the Cloud Run, Secret Manager and related APIs, and grants the default compute service account its roles. It pauses on errors and waits for input, so run it in a terminal: source init.sh If your session times out or you need to reset your variables, run set env.sh : source set env.sh Current Environment: GOOGLE CLOUD PROJECT=aisprint-491218 GOOGLE CLOUD LOCATION=us-east4 SERVICE NAME=gpu-2b-l4-devops-agent MODEL NAME=/mnt/models/gemma-4-E2B-it VLLM BASE URL=<unset — discovered via gcloud Cloud Run here is --no-allow-unauthenticated. If calls fail, run: source ./set adc.sh VLLM BASE URL can stay unset: the MCP server finds the service URL through gcloud when a tool first needs it. One of the key features that the MCP libraries provide is abstracting various transport methods. The tool implementation is the same no matter which transport the MCP client uses to connect. The simplest transport is stdio — the client launches the server as a local process and talks to it over stdin and stdout. Both must run in the same environment. In this project Claude Code is the MCP client. The server is created in one line: Initialize MCP server mcp 2.x; FastMCP was renamed MCPServer mcp = MCPServer "Self-Hosted vLLM DevOps Agent" That line used to say FastMCP . The next section is why it changed. Nothing in the repository changed. A fresh install did. requirements.txt listed mcp with no version bound, so the next pip install resolved the 2.x line, and the server stopped importing: python python3 -c "from mcp.server.fastmcp import FastMCP" raise ModuleNotFoundError MESSAGE, name= name ModuleNotFoundError: No module named 'mcp.server.fastmcp'. This is mcp 2.x, where FastMCP was renamed to MCPServer from mcp.server.mcpserver import MCPServer and other APIs changed; see the migration guide at https://py.sdk.modelcontextprotocol.io/v2/migration/ fastmcp-renamed-to-mcpserver or pin 'mcp<2' to keep running v1 code. In Claude Code it showed up less helpfully, as a server listed with Connection closed : it died on the import before the handshake. The error names two fixes. Pinning mcp<2 is legitimate — the v1 line still gets critical fixes — but these projects install into one system Python with no virtualenvs, so a pin here is a downgrade for every other project on the machine. Migrating keeps the change inside the repository. The whole code change is the import and the constructor: python -from mcp.server.fastmcp import FastMCP +from mcp.server.mcpserver import MCPServer ... -mcp = FastMCP "Self-Hosted vLLM DevOps Agent" +mcp = MCPServer "Self-Hosted vLLM DevOps Agent" @mcp.tool , @mcp.resource , mcp.run and every tool body stay as they are. Before editing, check the rest of the migration guide's list against your own server: grep -c "^@mcp\.\ tool\|resource\ " server.py grep -A1 "^@mcp\." server.py | grep -c "^def" grep -n "get running loop\|asyncio.run " server.py || echo " no matches " grep -n "^import httpx" server.py; grep -n "^httpx" requirements.txt python 28 12 no matches 13:import httpx 14:httpx Three things to know from that output: mcp 2.x no longer installs httpx . httpx2 instead. This server imports requirements.txt already declared it. version= to MCPServer ... . Nothing breaks; it shows in the handshake below. The requirement then says what the code needs — mcp =2 in place of the bare mcp line. This project has a Claude Code hook that runs ruff check --fix after every edit. Change the import line first and, for a moment, MCPServer is imported but unused — so the hook deletes it: python --- server.py +++ server.py @@ -1,3 +1,2 @@ -from mcp.server.mcpserver import MCPServer mcp = FastMCP "demo" Would fix 1 error. Make both changes in one edit, or change the usage first. Any editor that runs ruff check --fix on save does the same thing. The project can be linted: make lint ruff check . All checks passed ruff format --check . 14 files already formatted mypy . Success: no issues found in 6 source files and tested: make test ---------------------------------------------------------------------- Ran 28 tests in 1.151s OK The suite compares the registered tool set against a hard-coded list, so a tool that silently failed to register after the rename would fail here. ✅ Unit tests call Python. A client speaks JSON-RPC over stdio, so test that too. Hold stdin open with sleep — with a bare printf pipe the server sees end-of-input and exits after answering only initialize : { printf '%s\n' \ '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"probe","version":"0"}}}' \ '{"jsonrpc":"2.0","method":"notifications/initialized"}' \ '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}' \ '{"jsonrpc":"2.0","id":3,"method":"resources/list","params":{}}'; sleep 5; } \ | python3 server.py 2 /dev/null Summarised: initialize OK: name='Self-Hosted vLLM DevOps Agent' version='' proto 2025-06-18 tools/list OK: 27 tools - cloudrun analyze cloud logging, cloudrun analyze gpu logs, cloudrun check gpu quotas, cloudrun deploy, ... resources/list OK: 'config://vllm-deployment-template' 🟢 27 tools and the resource. Keep this snippet — it is the fastest way to tell "my server is broken" from "my client config is broken." Claude Code reads .mcp.json in the project directory. It launches server.py with the system python3 : { "mcpServers": { "cloudrun-devops": { "command": "python3", "args": "/home/xbill/gemma4-dev/gpu-2B-cloudrun-devops-agent/server.py" , "env": { "GOOGLE CLOUD PROJECT": "aisprint-491218", "GOOGLE CLOUD LOCATION": "us-east4", "VLLM BASE URL": "https://gpu-2b-l4-devops-agent-289270257791.us-east4.run.app", "MODEL NAME": "/mnt/models/gemma-4-E2B-it" } } } } Check the connection from Claude Code to the local server: claude mcp get cloudrun-devops cloudrun-devops: Scope: Project config shared via .mcp.json Status: ✔ Connected If the server failed at startup earlier in the session, reconnect it from /mcp or start a new session to pick up the fixed code. The MCP tools cover the whole lifecycle of the Cloud Run deployment. Every tool is prefixed cloudrun . Abridged output of cloudrun get help : The server is running in CLOUD RUN mode targeting NVIDIA L4 GPU in region us-east4. 🐳 Infrastructure & Deployment cloudrun deploy, cloudrun destroy, cloudrun status, cloudrun update scaling, cloudrun get deployment config, cloudrun get gpu deployment config, cloudrun check gpu quotas 📊 Model Management cloudrun list vertex models, cloudrun list bucket models, cloudrun save hf token, cloudrun get vertex ai model copy instructions, cloudrun get huggingface model copy instructions, cloudrun get huggingfacehub download path 📊 Monitoring & Status cloudrun get metrics, cloudrun get system status, cloudrun get endpoint, cloudrun get endpoint url, cloudrun get model details, cloudrun verify model health 📈 Performance & Benchmarking cloudrun run benchmark 💬 Interaction & Diagnostics cloudrun query gemma4, cloudrun query gemma4 with stats, cloudrun query, cloudrun analyze cloud logging, cloudrun analyze gpu logs, cloudrun suggest sre remediation Cloud Run mounts the bucket read-only at /mnt/models through GCS FUSE, so the weights go to GCS once. Download to a real disk rather than /tmp , which on this host is a RAM-backed tmpfs smaller than the model: hf download google/gemma-4-E2B-it --local-dir ~/hf-downloads/gemma-4-E2B-it gcloud storage rsync ~/hf-downloads/gemma-4-E2B-it gs://aisprint-491218-bucket/gemma-4-E2B-it \ --recursive --exclude='^\.cache/' gcloud storage ls -l gs://aisprint-491218-bucket/gemma-4-E2B-it/ Average throughput: 41.4MiB/s 4954 2026-09-10T14:51:14Z gs://aisprint-491218-bucket/gemma-4-E2B-it/config.json 10246621918 2026-09-10T14:55:13Z gs://aisprint-491218-bucket/gemma-4-E2B-it/model.safetensors 32169626 2026-09-10T14:51:34Z gs://aisprint-491218-bucket/gemma-4-E2B-it/tokenizer.json TOTAL: 9 objects, 10278849571 bytes 9.57GiB The bucket is shared with other models, and cloudrun list bucket models shows the trap: cloudrun list bucket models Contents of GCS Bucket: aisprint-491218-bucket - gemma-2b-it/config.json 0.00 MB - gemma-2b-it/model-00001-of-00002.safetensors 4716.15 MB ... - gemma-4-12B-it-qat-w4a16-ct/model.safetensors 9788.73 MB ... Check the architecture, not the folder name. gemma-2b-it/ looks like the answer and is the original Gemma, which the gemma4 parsers cannot serve. config.json settles it: gcloud storage cat gs://aisprint-491218-bucket/gemma-4-E2B-it/config.json \ | python3 -c 'import json,sys; c=json.load sys.stdin ; t=c "text config" ; print c "model type" , c "architectures" , "hidden", t "hidden size" , "layers", t "num hidden layers" ' gemma4 'Gemma4ForConditionalGeneration' hidden 1536 layers 35 The deploy-vllm target in the Makefile is the single source of truth for the vLLM and Cloud Run flags. The ones that matter most: | Flag | Value | Why | |---|---|---| | --gpu-type | nvidia-l4 | one L4 per instance | | --no-gpu-zonal-redundancy | | the cheaper L4 SKU | | --concurrency | 4 | requests Cloud Run sends one instance | | --max-num-seqs | 8 | vLLM's batch ceiling | | --tool-call-parser , --reasoning-parser | gemma4 | Gemma 4 tool calling breaks without either | The service is also --no-allow-unauthenticated , so every caller needs an identity token. make deploy Deploying container to Cloud Run service gpu-2b-l4-devops-agent in project aisprint-491218 region us-east4 Deploying new service... Routing traffic.....done Done. Service gpu-2b-l4-devops-agent revision gpu-2b-l4-devops-agent-00001-ssq has been deployed and is serving 100 percent of traffic. Service URL: https://gpu-2b-l4-devops-agent-289270257791.us-east4.run.app gcloud run deploy returns only once the startup probe passes, and the probe waits initialDelaySeconds=180 before its first check. Expect several minutes. The default autoscales between zero and one instance, which means a cold GPU start after the service goes idle. A demo cannot wait for that. The Makefile takes a SCALING variable: SCALING ?= auto ifeq $ SCALING ,auto SCALING FLAGS = --scaling=auto --max-instances=1 --min-instances=0 else SCALING FLAGS = --scaling=$ SCALING endif make deploy SCALING=1 gcloud run services describe gpu-2b-l4-devops-agent --region us-east4 --format='yaml metadata.annotations ' run.googleapis.com/manualInstanceCount: '1' run.googleapis.com/scalingMode: manual One L4 now runs until you change it. Plain make deploy , and the cloudrun deploy and cloudrun update scaling tools, all pass --scaling=auto on purpose — a service stuck in manual scaling at zero returns 503 to every request. So any of them quietly takes a demo back to scale-to-zero. The status can be checked with an MCP tool: cloudrun get system status 🌀 GPU Cloud Run System Status gpu-2b-l4-devops-agent - vLLM Health: 🟢 Online https://gpu-2b-l4-devops-agent-289270257791.us-east4.run.app - Cloud Run Service Status: 🟢 Ready 👉 Next Step: Use cloudrun query gemma4 to interact with the model. Ask vLLM what it loaded: curl -s -H "Authorization: Bearer $ gcloud auth print-identity-token " \ https://gpu-2b-l4-devops-agent-289270257791.us-east4.run.app/v1/models | python3 -m json.tool { "object": "list", "data": { "id": "/mnt/models/gemma-4-E2B-it", "object": "model", "owned by": "vllm", "max model len": 16384 } } The model id is the mount path, not the Hugging Face repo id. The container is started with --model=/mnt/models/<path , so that path is the name the OpenAI API expects. Then the MCP health check: cloudrun verify model health ✅ Model health check PASSED. Model: /mnt/models/gemma-4-E2B-it Response: 'Hello Yes, I am working. I am Gemma 4, a Large La...' Latency: 0.87 seconds. Right after the first deploy the same check took 2.48 seconds. That answer came through the whole stack: Claude Code called the tool over MCP, and the tool called vLLM on Cloud Run. 🟢 Ask the agent for cloudrun run benchmark with its defaults: one warmup request, then 20 requests at each concurrency level of 1, 2, 4 and 8, up to 128 output tokens each, one fixed prompt at temperature 0. | Concurrency | Req/s | Tokens/s | Avg latency | P95 latency | |---|---|---|---|---| | 1 | 0.39 | 49.63 | 2.58 s | 2.59 s | | 2 | 0.75 | 95.36 | 2.68 s | 2.74 s | | 4 | 1.47 | 188.46 | 2.71 s | 2.78 s | | 8 | 1.48 | 189.53 | 4.86 s | 5.47 s | Every request at every level succeeded. Three readings: From 1 to 4, throughput scales almost linearly. 188.46 tokens/s is 3.8x the single-stream 49.63 arithmetic , while average latency moves from 2.58 s to 2.71 s. The L4 is nowhere near full at 4. From 4 to 8, it stops. Throughput rises 0.6% arithmetic while average latency goes from 2.71 s to 4.86 s. Half the requests are waiting. The ceiling is a Cloud Run setting, not the GPU. One instance accepts --concurrency=4 requests at a time, and vLLM would batch up to --max-num-seqs=8 . The next section checks that against the same chip with nothing in front of it. The same checkpoint was served on the same chip — one NVIDIA L4 in an AWS EC2 g6.2xlarge — with vLLM and --max-num-seqs 8 , and swept with vllm bench serve at the same concurrency levels and 128 output tokens: | Concurrency | Cloud Run L4 tok/s | EC2 g6 L4 tok/s | | |---|---|---|---| | 1 | 49.63 | 46.09 | 🥇 Cloud Run | | 2 | 95.36 | 92.6 | 🥇 Cloud Run | | 4 | 188.46 | 175.83 | 🥇 Cloud Run | | 8 | 189.53 | 360.17 | 🥇 EC2 g6 | Up to 4, the two are the same chip doing the same work — within 8% of each other at every level, with Cloud Run slightly ahead. At 8, the g6 keeps going. It reaches 360.17 tokens/s, 1.9x Cloud Run's 189.53 arithmetic , because nothing in front of vLLM caps admissions at 4. That is the direct evidence that Cloud Run's plateau is --concurrency=4 , not the L4. Read the rows as a shape, not a leaderboard. The runs differ in engine version, KV-cache dtype, prompt and client location; the scope paragraph at the end lists each. Cloud Run bills a GPU service by the second for the whole instance while it runs — GPU, CPU and memory — because the L4 requires --no-cpu-throttling . List prices in us-east4 from the Cloud Billing catalog: | Component | Price | Per hour | |---|---|---| | NVIDIA L4, no zonal redundancy | $0.0001867 / s | $0.6721 | | 8 vCPU, instance-based | $0.000018 / vCPU-s | $0.5184 | | 32 GiB memory, instance-based | $0.000002 / GiB-s | $0.2304 | | One instance | | $1.4209 | That is $34.10 a day in demo mode with one fixed instance arithmetic . In the default mode an idle service scales to zero instances. Per million output tokens, from the benchmark arithmetic : | Deployment | Tokens/s | $ / hour | $ / M tokens | |---|---|---|---| | Cloud Run, concurrency 1 | 49.63 | 1.4209 | 7.95 | | Cloud Run, concurrency 4 | 188.46 | 1.4209 | 2.09 | | Cloud Run, concurrency 8 | 189.53 | 1.4209 | 2.08 | | 🥇 EC2 g6 spot, concurrency 8 | 360.17 | 0.9412 | 0.73 | The winner is the VM, on two counts: a lower hourly rate, and twice the throughput with no admission cap. Cloud Run's price is list and on-demand; the g6 price is a spot price and can be reclaimed. What Cloud Run buys is no VM to manage and zero cost while idle — for intermittent SRE work that is the number that matters. make destroy Not run for this article — the demo service is still up. It deletes the Cloud Run service; the weights stay in the bucket for the next deploy. The goal of this article was to deploy Gemma 4 E2B to a Cloud Run NVIDIA L4 GPU with vLLM, and to manage the whole lifecycle from Claude Code through a Python MCP server. The key to the solution was bringing the MCP server up and validating it locally before the deploy — which is where the move to the MCP SDK 2.x surfaced, and where it cost one import and one class name. The deployment results were: make deploy and passed the MCP health check in 0.87 seconds --concurrency=4 is the ceiling — the same L4 on EC2 reached 360.17 tokens/s at 8 Scope: one Cloud Run instance with one NVIDIA L4 in us-east4 , vLLM v0.26.0-cu129 , Gemma 4 E2B with bf16 weights and fp8 KV cache, manual scaling at one fixed instance during the sweep. One sweep of 20 requests per level, a single fixed prompt, 128 max output tokens, driven over the internet from a laptop. The EC2 comparison is a separate run: vLLM 0.28.0, bf16 KV cache, 1,024-token random prompts, 8 to 32 requests per level, driven on the instance itself, spot-priced in us-east-1 . Costs are arithmetic on list prices, not a bill. MCP server on mcp 2.2.0 and Python 3.14.7. The strategy for using MCP for Gemma 4 GPU deployment to Cloud Run with Claude Code was validated with an incremental step by step approach.