[GH-ISSUE #91 ] JSONDecodeError: Expecting value: line 5 column 1 (char 4) #77

Author

Owner

@pedayako commented on GitHub (Jan 28, 2025):

I'm pretty sure. I think the best solution is to just wait until the services are working well again.

@pedayako commented on GitHub (Jan 28, 2025): I'm pretty sure. I think the best solution is to just wait until the services are working well again.

bsakaguchi commented

Author

Owner

@zhuofengli commented on GitHub (Jan 28, 2025):

We already use max_token as part of our api submit - makes no difference we get the " Expecting value: line 5 column 1 (char 4)" error on api calls both to V3 and R1. That said, platform api for both models worked perfectly fine until Sunday night (before media hysteria started), so hoping this is a ddos related issue and will resolve.

When I set max_token to a low number like 300, it works, but not for a high number like 4000

@zhuofengli commented on GitHub (Jan 28, 2025): > We already use `max_token` as part of our api submit - makes no difference we get the " Expecting value: line 5 column 1 (char 4)" error on api calls both to V3 and R1. That said, platform api for both models worked perfectly fine until Sunday night (before media hysteria started), so hoping this is a ddos related issue and will resolve. When I set max_token to a low number like 300, it works, but not for a high number like 4000

bsakaguchi commented

Author

Owner

@devanshg03 commented on GitHub (Jan 29, 2025):

I seem to have the same issue regardless of my setting for max_tokens or even if I exclude it. Even using their official python example below. I wonder if there isn't proper error handling on the server side, and thus, it doesn't decode properly when there is an error?

from dotenv import load_dotenv
import os

from openai import OpenAI
load_dotenv()

client = OpenAI(api_key=os.getenv("DEEPSEEK_API_KEY"), base_url="https://api.deepseek.com")

response = client.chat.completions.create(
    model="deepseek-reasoner",
    messages=[
        {"role": "system", "content": "You are a helpful assistant"},
        {"role": "user", "content": "Hello"},
    ],
    stream=False,
    max_token=300
)

print(response.choices[0].message.content)

@devanshg03 commented on GitHub (Jan 29, 2025): I seem to have the same issue regardless of my setting for `max_tokens` or even if I exclude it. Even using their official python example below. I wonder if there isn't proper error handling on the server side, and thus, it doesn't decode properly when there is an error? ``` from dotenv import load_dotenv import os from openai import OpenAI load_dotenv() client = OpenAI(api_key=os.getenv("DEEPSEEK_API_KEY"), base_url="https://api.deepseek.com") response = client.chat.completions.create( model="deepseek-reasoner", messages=[ {"role": "system", "content": "You are a helpful assistant"}, {"role": "user", "content": "Hello"}, ], stream=False, max_token=300 ) print(response.choices[0].message.content) ```

bsakaguchi commented

2026-10-26 21:13:16 +00:00

Author

Owner

@pedayako commented on GitHub (Jan 29, 2025):

I think there's nothing wrong with your code. The model's reasoning seems to have an issue since the attacks. If you try with the chat model, it will probably work well.

@pedayako commented on GitHub (Jan 29, 2025): I think there's nothing wrong with your code. The model's reasoning seems to have an issue since the attacks. If you try with the chat model, it will probably work well.

bsakaguchi commented

Author

Owner

@globalcargo commented on GitHub (Jan 30, 2025):

As of right now both models (V3 and R-1) seems to be back online again and working via the platform API. No more JSONDecodeError. We did not modify our code. So this seems to confirm what others have posited, that this was an issue related to the attacks. Feel free to close this issue for now.

@globalcargo commented on GitHub (Jan 30, 2025): As of right now both models (V3 and R-1) seems to be back online again and working via the platform API. No more JSONDecodeError. We did not modify our code. So this seems to confirm what others have posited, that this was an issue related to the attacks. Feel free to close this issue for now.

bsakaguchi commented

2026-10-26 21:13:16 +00:00

Author

Owner

@zhangzx-uiuc commented on GitHub (Jan 30, 2025):

I still get JSONDecodeError occasionally..

@zhangzx-uiuc commented on GitHub (Jan 30, 2025): I still get JSONDecodeError occasionally..

bsakaguchi commented

2026-10-26 21:13:16 +00:00

Author

Owner

@lsjlsj5846 commented on GitHub (Jan 31, 2025):

I also get JSONDecodeError now; setting max_tokens does not help fixing this issue.

@lsjlsj5846 commented on GitHub (Jan 31, 2025): I also get JSONDecodeError now; setting `max_tokens` does not help fixing this issue.

bsakaguchi commented

2026-10-26 21:13:17 +00:00

Author

Owner

@kenshiro-o commented on GitHub (Feb 4, 2025):

Same issue here. It may be due to the unexpected high load their servers have been under for a while...
Worth checking again in a few days

@kenshiro-o commented on GitHub (Feb 4, 2025): Same issue here. It may be due to the unexpected high load their servers have been under for a while... Worth checking again in a few days

bsakaguchi commented

Author

Owner

@reyynaldi commented on GitHub (Feb 5, 2025):

its kinda weird to say but im glad everyone got the same problem too 😭 i thought its just me. Ya i hope everythings going to work like they used to be again. Anyway, ive tried using openrouter api for deepseek (both reasoner and chat model), while it works, its annoyingly slow, like one simple prompt could take like 40s second. crazy.

@reyynaldi commented on GitHub (Feb 5, 2025): its kinda weird to say but im glad everyone got the same problem too 😭 i thought its just me. Ya i hope everythings going to work like they used to be again. Anyway, ive tried using openrouter api for deepseek (both reasoner and chat model), while it works, its annoyingly slow, like one simple prompt could take like 40s second. crazy.

bsakaguchi commented

Author

Owner

@liuhanzuo commented on GitHub (Feb 6, 2025):

I think that the problem is basically caused by the api platform. However, add a max_tokens indeed helps, but can not prevent the problem of connection error. Since the connection error is not that frequently, I think retry for several times would help

@liuhanzuo commented on GitHub (Feb 6, 2025): I think that the problem is basically caused by the api platform. However, add a max_tokens indeed helps, but can not prevent the problem of connection error. Since the connection error is not that frequently, I think retry for several times would help

bsakaguchi commented

Author

Owner

@mc-brgb commented on GitHub (Feb 6, 2025):

Same issue here, first query would usually work then consistently get the JSON decoder error with both chat and reasoner model. It was working fine before the media attention.

@mc-brgb commented on GitHub (Feb 6, 2025): Same issue here, first query would usually work then consistently get the JSON decoder error with both chat and reasoner model. It was working fine before the media attention.

bsakaguchi commented

Author

Owner

@globalcargo commented on GitHub (Feb 6, 2025):

We are observing the same thing whereby as the conversation history increases and thus input tokens increase, eventually after two or three turns the model will fail with the JSON decoder error. This happens for both the regular V3 as well as the R1 model.

@globalcargo commented on GitHub (Feb 6, 2025): We are observing the same thing whereby as the conversation history increases and thus input tokens increase, eventually after two or three turns the model will fail with the JSON decoder error. This happens for both the regular V3 as well as the R1 model.

bsakaguchi commented

Author

Owner

@Junjie-Chu commented on GitHub (Feb 7, 2025):

Same problems here. I guess this is something due to high workload of their servers.

@Junjie-Chu commented on GitHub (Feb 7, 2025): Same problems here. I guess this is something due to high workload of their servers.

bsakaguchi commented

Author

Owner

@pranavmalikk commented on GitHub (Feb 7, 2025):

Same problems here

@pranavmalikk commented on GitHub (Feb 7, 2025): Same problems here

bsakaguchi commented

Author

Owner

@agoyang commented on GitHub (Feb 7, 2025):

I have encountered this problem for several consecutive days. It seems that the first few requests every day are fine, and then this error is reported all the time.

@agoyang commented on GitHub (Feb 7, 2025): I have encountered this problem for several consecutive days. It seems that the first few requests every day are fine, and then this error is reported all the time.

bsakaguchi commented

Author

Owner

@zyuanat commented on GitHub (Feb 7, 2025):

response = client.chat.completions.create(
model="deepseek-chat",
messages=[
{"role": "system", "content": "You are a helpful assistant"},
{"role": "user", "content": "Hello"},
],
stream=False
)

The same problem persisted for days, with even the example request raising "raise JSONDecodeError("Expecting value", s, err.value) from None
json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)"

Occasionally it's normal, but most of the time it continues to report errors like this. Especially with multiple calls.

@zyuanat commented on GitHub (Feb 7, 2025): response = client.chat.completions.create( model="deepseek-chat", messages=[ {"role": "system", "content": "You are a helpful assistant"}, {"role": "user", "content": "Hello"}, ], stream=False ) The same problem persisted for days, with even the example request raising "raise JSONDecodeError("Expecting value", s, err.value) from None json.decoder.JSONDecodeError: Expecting value: line 1 column 1 (char 0)" Occasionally it's normal, but most of the time it continues to report errors like this. Especially with multiple calls.

bsakaguchi commented

Author

Owner

@crescentluna commented on GitHub (Feb 7, 2025):

same problems, still don't work

@crescentluna commented on GitHub (Feb 7, 2025): same problems, still don't work

bsakaguchi commented

Author

Owner

@hpclass commented on GitHub (Feb 7, 2025):

我也遇到同样的问题，使用https://api.deepseek.com/v1进行请求的时候间歇性抛出”line 1 column 1 (char 0)“，也有能正常返回结果的时候，使用第三方代理API则正常，怀疑是服务器有问题。。。

@hpclass commented on GitHub (Feb 7, 2025): 我也遇到同样的问题，使用https://api.deepseek.com/v1进行请求的时候间歇性抛出”line 1 column 1 (char 0)“，也有能正常返回结果的时候，使用第三方代理API则正常，怀疑是服务器有问题。。。

bsakaguchi commented

Author

Owner

@yulijia commented on GitHub (Feb 7, 2025):

I am an international user, I got the line 1 column 1 (char 0) error every minute today.

@yulijia commented on GitHub (Feb 7, 2025): I am an international user, I got the line 1 column 1 (char 0) error every minute today.

bsakaguchi commented

Author

Owner

@baocai27 commented on GitHub (Feb 8, 2025):

JSONDecodeError: Expecting value: line 1 column 1 (char 0)
same problem QAQ

@baocai27 commented on GitHub (Feb 8, 2025): JSONDecodeError: Expecting value: line 1 column 1 (char 0) same problem QAQ

bsakaguchi commented