[Bug]: GPT-OSS-120B returns null content when hitting max_tokens without --enforce-eager

[ashtarkb]

Your current environment

The output of python collect_env.py

==============================
        System Info
==============================
OS                           : Ubuntu 22.04.5 LTS (x86_64)
GCC version                  : (Ubuntu 11.4.0-1ubuntu1~22.04.2) 11.4.0
Clang version                : Could not collect
CMake version                : Could not collect
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.9.0+cu129
Is debug build               : False
CUDA used to build PyTorch   : 12.9
ROCM used to build PyTorch   : N/A

==============================
      Python Environment
==============================
Python version               : 3.12.12 (main, Oct 10 2025, 08:52:57) [GCC 11.4.0] (64-bit runtime)
Python platform              : Linux-5.14.0-427.65.1.el9_4.x86_64-x86_64-with-glibc2.35

==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.9.86
CUDA_MODULE_LOADING set to   :
GPU models and configuration :
GPU 0: NVIDIA A100-SXM4-80GB
GPU 1: NVIDIA A100-SXM4-80GB
GPU 2: NVIDIA A100-SXM4-80GB
GPU 3: NVIDIA A100-SXM4-80GB
GPU 4: NVIDIA A100-SXM4-80GB
GPU 5: NVIDIA A100-SXM4-80GB
GPU 6: NVIDIA A100-SXM4-80GB
GPU 7: NVIDIA A100-SXM4-80GB

Nvidia driver version        : 550.163.01
cuDNN version                : Could not collect
HIP runtime version          : N/A
MIOpen runtime version       : N/A
Is XNNPACK available         : True

==============================
          CPU Info
==============================
Architecture:                         x86_64
CPU op-mode(s):                       32-bit, 64-bit
Address sizes:                        46 bits physical, 57 bits virtual
Byte Order:                           Little Endian
CPU(s):                               80
On-line CPU(s) list:                  0-79
Vendor ID:                            GenuineIntel
Model name:                           Intel Xeon Processor (Icelake)
CPU family:                           6
Model:                                134
Thread(s) per core:                   2
Core(s) per socket:                   20
Socket(s):                            2
Stepping:                             0
BogoMIPS:                             5600.01
Flags:                                fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc rep_good nopl xtopology cpuid tsc_known_freq pni pclmulqdq vmx ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch cpuid_fault ssbd ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves wbnoinvd arat vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid fsrm md_clear arch_capabilities
Virtualization:                       VT-x
Hypervisor vendor:                    KVM
Virtualization type:                  full
L1d cache:                            2.5 MiB (80 instances)
L1i cache:                            2.5 MiB (80 instances)
L2 cache:                             160 MiB (40 instances)
L3 cache:                             32 MiB (2 instances)
NUMA node(s):                         2
NUMA node0 CPU(s):                    0-39
NUMA node1 CPU(s):                    40-79
Vulnerability Gather data sampling:   Not affected
Vulnerability Itlb multihit:          Not affected
Vulnerability L1tf:                   Not affected
Vulnerability Mds:                    Not affected
Vulnerability Meltdown:               Not affected
Vulnerability Mmio stale data:        Vulnerable: Clear CPU buffers attempted, no microcode; SMT Host state unknown
Vulnerability Reg file data sampling: Vulnerable: No microcode
Vulnerability Retbleed:               Not affected
Vulnerability Spec rstack overflow:   Not affected
Vulnerability Spec store bypass:      Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:             Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:             Mitigation; Enhanced / Automatic IBRS; IBPB conditional; RSB filling; PBRSB-eIBRS Not affected; BHI SW loop, KVM SW loop
Vulnerability Srbds:                  Not affected
Vulnerability Tsx async abort:        Not affected

==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.5.3
[pip3] numpy==2.2.0
[pip3] nvidia-cublas-cu12==12.9.1.4
[pip3] nvidia-cuda-cupti-cu12==12.9.79
[pip3] nvidia-cuda-nvrtc-cu12==12.9.86
[pip3] nvidia-cuda-runtime-cu12==12.9.79
[pip3] nvidia-cudnn-cu12==9.10.2.21
[pip3] nvidia-cudnn-frontend==1.16.0
[pip3] nvidia-cufft-cu12==11.4.1.4
[pip3] nvidia-cufile-cu12==1.14.1.1
[pip3] nvidia-curand-cu12==10.3.10.19
[pip3] nvidia-cusolver-cu12==11.7.5.82
[pip3] nvidia-cusparse-cu12==12.5.10.65
[pip3] nvidia-cusparselt-cu12==0.7.1
[pip3] nvidia-cutlass-dsl==4.3.1
[pip3] nvidia-ml-py==13.580.82
[pip3] nvidia-nccl-cu12==2.27.5
[pip3] nvidia-nvjitlink-cu12==12.9.86
[pip3] nvidia-nvshmem-cu12==3.3.20
[pip3] nvidia-nvtx-cu12==12.9.79
[pip3] pyzmq==27.1.0
[pip3] torch==2.9.0+cu129
[pip3] torchaudio==2.9.0+cu129
[pip3] torchvision==0.24.0+cu129
[pip3] transformers==4.57.3
[pip3] triton==3.5.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
vLLM Version                 : 0.12.0
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled
GPU Topology:
  	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	NIC0	CPU Affinity	NUMA Affinity	GPU NUMA ID
GPU0	 X 	NV12	NV12	NV12	NV12	NV12	NV12	NV12	PIX	0-39	0		N/A
GPU1	NV12	 X 	NV12	NV12	NV12	NV12	NV12	NV12	PIX	0-39	0		N/A
GPU2	NV12	NV12	 X 	NV12	NV12	NV12	NV12	NV12	NODE	0-39	0		N/A
GPU3	NV12	NV12	NV12	 X 	NV12	NV12	NV12	NV12	NODE	0-39	0		N/A
GPU4	NV12	NV12	NV12	NV12	 X 	NV12	NV12	NV12	SYS	40-79	1		N/A
GPU5	NV12	NV12	NV12	NV12	NV12	 X 	NV12	NV12	SYS	40-79	1		N/A
GPU6	NV12	NV12	NV12	NV12	NV12	NV12	 X 	NV12	SYS	40-79	1		N/A
GPU7	NV12	NV12	NV12	NV12	NV12	NV12	NV12	 X 	SYS	40-79	1		N/A
NIC0	PIX	PIX	NODE	NODE	SYS	SYS	SYS	SYS	 X

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

NIC Legend:

  NIC0: mlx5_0

==============================
     Environment Variables
==============================
VLLM_USAGE_SOURCE=production-docker-image
LD_LIBRARY_PATH=/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib64
NVIDIA_DRIVER_CAPABILITIES=compute,utility
NVIDIA_REQUIRE_CUDA=cuda>=12.9 brand=unknown,driver>=535,driver<536 brand=grid,driver>=535,driver<536 brand=tesla,driver>=535,driver<536 brand=nvidia,driver>=535,driver<536 brand=quadro,driver>=535,driver<536 brand=quadrortx,driver>=535,driver<536 brand=nvidiartx,driver>=535,driver<536 brand=vapps,driver>=535,driver<536 brand=vpc,driver>=535,driver<536 brand=vcs,driver>=535,driver<536 brand=vws,driver>=535,driver<536 brand=cloudgaming,driver>=535,driver<536 brand=unknown,driver>=550,driver<551 brand=grid,driver>=550,driver<551 brand=tesla,driver>=550,driver<551 brand=nvidia,driver>=550,driver<551 brand=quadro,driver>=550,driver<551 brand=quadrortx,driver>=550,driver<551 brand=nvidiartx,driver>=550,driver<551 brand=vapps,driver>=550,driver<551 brand=vpc,driver>=550,driver<551 brand=vcs,driver>=550,driver<551 brand=vws,driver>=550,driver<551 brand=cloudgaming,driver>=550,driver<551 brand=unknown,driver>=560,driver<561 brand=grid,driver>=560,driver<561 brand=tesla,driver>=560,driver<561 brand=nvidia,driver>=560,driver<561 brand=quadro,driver>=560,driver<561 brand=quadrortx,driver>=560,driver<561 brand=nvidiartx,driver>=560,driver<561 brand=vapps,driver>=560,driver<561 brand=vpc,driver>=560,driver<561 brand=vcs,driver>=560,driver<561 brand=vws,driver>=560,driver<561 brand=cloudgaming,driver>=560,driver<561 brand=unknown,driver>=565,driver<566 brand=grid,driver>=565,driver<566 brand=tesla,driver>=565,driver<566 brand=nvidia,driver>=565,driver<566 brand=quadro,driver>=565,driver<566 brand=quadrortx,driver>=565,driver<566 brand=nvidiartx,driver>=565,driver<566 brand=vapps,driver>=565,driver<566 brand=vpc,driver>=565,driver<566 brand=vcs,driver>=565,driver<566 brand=vws,driver>=565,driver<566 brand=cloudgaming,driver>=565,driver<566 brand=unknown,driver>=570,driver<571 brand=grid,driver>=570,driver<571 brand=tesla,driver>=570,driver<571 brand=nvidia,driver>=570,driver<571 brand=quadro,driver>=570,driver<571 brand=quadrortx,driver>=570,driver<571 brand=nvidiartx,driver>=570,driver<571 brand=vapps,driver>=570,driver<571 brand=vpc,driver>=570,driver<571 brand=vcs,driver>=570,driver<571 brand=vws,driver>=570,driver<571 brand=cloudgaming,driver>=570,driver<571
NVIDIA_VISIBLE_DEVICES=all
CUDA_VERSION=12.9.1
CUDA_LAUNCH_BLOCKING=1
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1

🐛 Describe the bug

When running GPT-OSS-120B without the --enforce-eager flag, the model returns content: null when responses hit max_tokens, despite successfully generating tokens server-side and returning HTTP 200 OK.

Key characteristics:

• Non-deterministic initially: First 1-2 requests may succeed, then subsequent requests fail

• Becomes deterministic: After warm-up, fails consistently (100% failure rate)

• Silent failure: No errors thrown, just null content with 200 OK status

• Server-side generation succeeds: Logs show ~120-150 tokens/s throughput and 200 OK

• Client-side receives nothing: content: null, reasoning_content: ""

Minimal reproduction code:

import requests

# Run this 3-5 times - first may work, then consistently fails
response = requests.post(
    'http://localhost:8000/v1/chat/completions',
    json={
        'model': '/models/gpt-oss-120b',
        'messages': [
            {'role': 'system', 'content': 'You are a helpful assistant.'},
            {'role': 'user', 'content': '''Analyze this complex distributed system architecture with microservices for API gateway, authentication, user management, content processing, search, analytics, notifications, caching, message queues, and databases. The system must handle 100K requests/second with 99.99% uptime across three availability zones. Provide detailed recommendations for fault tolerance, service mesh optimization, monitoring strategies, disaster recovery with RTO<1hr and RPO<5min, CI/CD improvements, cost optimization (40% reduction), SLO definitions, and incident response procedures. Include specific technologies and implementation steps.'''}
        ],
        'max_tokens': 2048
    },
    timeout=120
)

result = response.json()
msg = result['choices'][0]['message']
usage = result['usage']

print(f"HTTP Status: {response.status_code}")
print(f"Content is None: {msg.get('content') is None}")
print(f"Tokens used: {usage['completion_tokens']}")
print(f"Finish reason: {result['choices'][0]['finish_reason']}")

Observed behavior:
HTTP Status: 200
Content is None:
TrueTokens used: 2048
Finish reason: length
Full response structure:

{  "choices": [{    "message": {      "role": "assistant",      "content": null,      "reasoning": "",      "reasoning_content": ""    },    "finish_reason": "length"  }],  "usage": {    "prompt_tokens": 512,    "completion_tokens": 2048  }}

Server logs during the failed request:
INFO: Avg generation throughput: 148.0 tokens/sINFO: Avg generation throughput: 114.8 tokens/sINFO: HTTP/1.1 200 OK
Tokens ARE generated server-side (confirmed by throughput metrics) but never appear in the API response.

Additional observations:

• Tested with different max_tokens values: 2048, 5000, 10000, 15000 - all fail when limit is hit
• Tested with different --max-model-len settings: 25000, 32768 - bug persists
• Simple prompts work: Short prompts like "Say hello" return content correctly
• Complex/long prompts fail: After 1-2 warm-up requests, consistently return null

Debug logs show CUDA graphs are active:
cudagraph_mode: FULL (across all 8 tensor parallel workers)

Workaround:
Adding --enforce-eager flag fixes the issue completely (100% success rate).
With this flag, the same prompts complete successfully with proper content formatting.

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.

[alew3]

I think it is the expected behavior, If you reach max tokens used in the reasoning you will get no tokens in the content output.

Try updating your code

result = response.json()
msg = result['choices'][0]['message']
usage = result['usage']

print(f"HTTP Status: {response.status_code}")
print(f"Finish reason: {result['choices'][0]['finish_reason']}")
print(f"Tokens used: {usage['completion_tokens']}")
print(f"Message keys: {list(msg.keys())}")

# Check both content and reasoning_content (for reasoning models)
content = msg.get('content')
reasoning_content = msg.get('reasoning_content')

print(f"\nContent is None: {content is None}")
print(f"Reasoning content is None: {reasoning_content is None}")

if content:
    print(f"\n--- Content (first 500 chars) ---\n{content[:500]}...")
if reasoning_content:
    print(f"\n--- Reasoning content (first 500 chars) ---\n{reasoning_content[:500]}...")

[bbrowning]

I suspect this is hitting the same infinite token id 0 generation bug reported in #26480, given that it fails after warm up (ie after hitting prefix cache), that --enforce-eager works around it (because that disables cudagraphs), and that you tested with max_tokens plenty high enough to give a reasonable answer.

[bbrowning]

I was able to confirm this is the same underlying issue as #26480, was able to reproduce it locally on gpt-oss-20b, and both of these issues are fixed for me by the small change in #30650. If you're able to test that pull request on your hardware with gpt-oss-120b, that would help confirm that this resolves the issue generally.

[bbrowning]

Note that in the reproducer python code shown originally for this issue, I need to set max_tokens to 10k+ as sometimes the model gets really verbose. But, for purposes of digging into this issue, I had a separate check in gpu_model_running.py to raise when it detects repeated NaN model_output generation so can definitively confirm this without relying just on the presence of content or not in this python script.