diff --git a/README.md b/README.md
index ce097f3..62407b3 100644
--- a/README.md
+++ b/README.md
@@ -136,6 +136,17 @@ let g:ollama_model = 'qwen2.5-coder:3b'
 
 " Deepseek-coder-v2
 let g:ollama_model = 'deepseek-coder-v2:16b-lite-base-q4_0'
+
+OpenAI-like FIM interface
+
+```
+let g:ollama_host = 'https://api.siliconflow.cn/v1'
+let g:ollama_model = 'deepseek-ai/DeepSeek-V2.5'
+let g:ollama_key = 'sk-xxxxxxxxxxxxx'
+let g:ollama_context_lines = 100
+```
+
+./autoload/ollama.vim
 ```
 
 | Variable              | Default                  | Description                            |
diff --git a/autoload/ollama.vim b/autoload/ollama.vim
index c1e8d09..a9c4645 100644
--- a/autoload/ollama.vim
+++ b/autoload/ollama.vim
@@ -107,7 +107,7 @@ function! ollama#GetSuggestion(timer)
     let s:timer_id = -1
     let l:current_line = line('.')
     let l:current_col = col('.')
-    let l:context_lines = 30
+    let l:context_lines = g:ollama_context_lines
 
     " Get the lines before and after the current line
     let l:prefix_lines = getline(max([1, l:current_line - l:context_lines]), l:current_line - 1)
@@ -129,7 +129,7 @@ function! ollama#GetSuggestion(timer)
     let l:prompt = l:prefix . '<FILL_IN_HERE>' . l:suffix
 
     let l:model_options = substitute(json_encode(g:ollama_model_options), "\"", "\\\"", "g")
-    call ollama#logger#Debug("Connecting to Ollama on ".g:ollama_host." using model ".g:ollama_model)
+    call ollama#logger#Debug("Connecting to Ollama on ".g:ollama_host." using model ".g:ollama_model." api_type".g:ollama_api_type)
     call ollama#logger#Debug("model_options=".l:model_options)
     " Convert plugin debug level to python logger levels
     let l:log_level = ollama#logger#PythonLogLevel(g:ollama_debug)
@@ -138,7 +138,9 @@ function! ollama#GetSuggestion(timer)
         \ "-m", g:ollama_model,
         \ "-u", g:ollama_host,
         \ "-o", l:model_options,
-        \ "-l", l:log_level
+        \ "-l", l:log_level,
+        \ "-t", g:ollama_api_type,
+        \ "-s", g:ollama_key,
         \ ]
     call ollama#logger#Debug("command=". join(l:command, " "))
     let l:job_options = {
diff --git a/plugin/ollama.vim b/plugin/ollama.vim
index cb5df22..9ed336f 100644
--- a/plugin/ollama.vim
+++ b/plugin/ollama.vim
@@ -40,6 +40,21 @@ if !exists('g:ollama_timeout')
     let g:ollama_timeout = 10
 endif
 
+if !exists('g:ollama_key')
+    " default code completion model
+    let g:ollama_key = ''
+endif
+
+if !exists('g:ollama_api_type')
+    " default to ollama
+    let g:ollama_api_type = 'ollama'
+endif
+
+if !exists('g:ollama_context_lines')
+    " default to 30
+    let g:ollama_context_lines = 30
+endif
+
 " Defines the color scheme for ollama suggestions
 function! s:ColorScheme() abort
     if &t_Co == 256
diff --git a/python/complete.py b/python/complete.py
index f302613..72d763d 100755
--- a/python/complete.py
+++ b/python/complete.py
@@ -10,7 +10,7 @@
 # Default values
 DEFAULT_HOST = 'http://localhost:11434'
 DEFAULT_MODEL = 'codellama:code'
-DEFAULT_OPTIONS = '{ "temperature": 0, "top_p": 0.95 }'
+DEFAULT_OPTIONS = '{ "temperature": 0, "top_p": 0.95}'
 
 # create logger
 log = OllamaLogger('ollama.log')
@@ -93,12 +93,45 @@ def generate_code_completion(config, prompt, baseurl, model, options):
     else:
         raise Exception(f"Error: {response.status_code} - {response.text}")
 
+def generate_code_completion_openai(prompt, url, key, model, options, max_tokens=256):
+    parts = prompt.split('<FILL_IN_HERE>')
+
+    if len(parts) != 2:
+        log.error("Prompt does not contain '<FILL_IN_HERE>'.")
+        sys.exit(1)
+
+    newprompt = parts[0]
+    new_suffix = parts[1]
+    log.debug(f"newprompt: {newprompt}, new_suffix: {new_suffix}")
+
+    from openai import OpenAI
+    client = OpenAI(api_key=key, base_url=url)
+
+    try:
+        response = client.completions.create(
+                model=model,
+                prompt=newprompt,
+                suffix=new_suffix,
+                stream=False,
+                max_tokens=max_tokens,
+                **options
+
+        )
+
+        text = response.choices[0].text.strip()
+        return text
+    except Exception as e:
+        log.error(f"Error: {e}")
+        sys.exit(1)
+
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="Complete code with Ollama LLM.")
     parser.add_argument('-m', '--model', type=str, default=DEFAULT_MODEL, help="Specify the model name to use.")
     parser.add_argument('-u', '--url', type=str, default=DEFAULT_HOST, help="Specify the base endpoint URL to use (default="+DEFAULT_HOST+")")
     parser.add_argument('-o', '--options', type=str, default=DEFAULT_OPTIONS, help="Specify the Ollama REST API options.")
     parser.add_argument('-l', '--log-level', type=int, default=OllamaLogger.ERROR, help="Specify log level")
+    parser.add_argument('-t', '--api-type', type=str, default='ollama', help="ollama or openai")
+    parser.add_argument('-s', '--key', type=str, default='', help="please specify the key")
     args = parser.parse_args()
 
     log.setLevel(args.log_level)
@@ -112,8 +145,14 @@ def generate_code_completion(config, prompt, baseurl, model, options):
     # strip suffix (e.g ':7b-code') from modelname
     modelname = args.model
     modelname = modelname.rsplit(':', 1)[0]
-    config = load_config(modelname)
 
     prompt = sys.stdin.read()
-    response = generate_code_completion(config, prompt, args.url, args.model, options)
+    if args.api_type == 'ollama':
+        config = load_config(modelname)
+        response = generate_code_completion(config, prompt, args.url, args.model, options)
+    elif args.api_type == 'openai':
+        response = generate_code_completion_openai(prompt, args.url, args.key, args.model, options)
+    else:
+        log.error("Please specify the correct api type")
+        sys.exit(1)
     print(response, end='')