Your current environment

============================== System Info

OS : Ubuntu 24.04.2 LTS (x86_64) GCC version : (Ubuntu 9.5.0-6ubuntu2) 9.5.0 Clang version : Could not collect CMake version : version 3.28.3 Libc version : glibc-2.39

============================== vLLM Info

ROCM Version : Could not collect vLLM Version : 0.19.0 vLLM Build Flags: CUDA Archs: Not Set; ROCm: Disabled; XPU: Disabled GPU Topology: GPU0 GPU1 CPU Affinity NUMA Affinity GPU NUMA ID GPU0 X PHB 0-7 0 N/A GPU1 PHB X 0-7 0 N/A

Legend:

X = Self SYS = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI) NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node PHB = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU) PXB = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge) PIX = Connection traversing at most a single PCIe bridge NV# = Connection traversing a bonded set of # NVLinks

🐛 Describe the bug

Bug Summary

In vLLM 0.19.1, the Gemma4ToolParser produces incorrect floating-point argument values in streaming mode. A number like 108.2 is incorrectly streamed as 108.02, and 22.8 becomes 22.08. String-typed arguments and non-streaming responses are unaffected.

Launch Command

vllm serve /llm-models/gemma-4-26B-A4B-it \
  --served-model-name gemma4-26b-a4b-it \
  --host=0.0.0.0 \
  --port 9999 \
  --gpu_memory_utilization 0.9 \
  --tensor-parallel-size 2 \
  --max-model-len 262144 \
  --enable-prefix-caching \
  --enable-auto-tool-choice \
  --reasoning-parser gemma4 \
  --tool-call-parser gemma4

Minimal Reproduction

# pip install openai==1.76.1
from openai import OpenAI

client = OpenAI(base_url="http://127.0.0.1:9999/v1", api_key="EMPTY")

response = client.chat.completions.create(
​    model="gemma4-26b-a4b-it",
​    messages=[{
​        "role": "user",
​        "content": [{"type": "text", "text": "Calculate the sum of 108.2 and 22.8"}]
​    }],
​    tools=[{
​        "type": "function",
​        "function": {
​            "name": "add",
​            "description": "Calculate the sum of two numbers",
​            "parameters": {
​                "type": "object",
​                "properties": {
​                    "left": {"type": "number", "description": "The first number"},
​                    "right": {"type": "number", "description": "The second number"}
​                },
​                "required": ["left", "right"]
​            }
​        }
​    }],
​    stream=True,
​    temperature=1.0,
​    top_p=0.95,
​    presence_penalty=1.5,
​    extra_body={
​        "top_k": 20,
​        "chat_template_kwargs": {"enable_thinking": False},
​    },
)

arguments = ""
for chunk in response:
​    tc = chunk.choices[0].delta.tool_calls
​    if tc and len(tc) == 1:
​        arguments += tc[0].function.arguments

print(arguments)
# Actual (streaming):   {"left": 108.02, "right": 22.08}
# Expected (streaming): {"left": 108.2,  "right": 22.8}

Expected vs Actual Behavior

Expected (streaming): {"left": 108.2, "right": 22.8} Actual (streaming): {"left": 108.02, "right": 22.08}

Before submitting a new issue...

• Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.