Merge pull request #2535 from BerriAI/litellm_fireworks_ai_support

feat(utils.py): add native fireworks ai support

docs/my-website/docs/providers/fireworks_ai.md

@@ -0,0 +1,53 @@
+# Fireworks AI
+https://fireworks.ai/
+
+**We support ALL Fireworks AI models, just set `fireworks_ai/` as a prefix when sending completion requests**
+
+## API Key
+```python
+# env variable
+os.environ['FIREWORKS_AI_API_KEY']
+```
+
+## Sample Usage
+```python
+from litellm import completion
+import os
+
+os.environ['FIREWORKS_AI_API_KEY'] = ""
+response = completion(
+    model="fireworks_ai/mixtral-8x7b-instruct", 
+    messages=[
+       {"role": "user", "content": "hello from litellm"}
+   ],
+)
+print(response)
+```
+
+## Sample Usage - Streaming
+```python
+from litellm import completion
+import os
+
+os.environ['FIREWORKS_AI_API_KEY'] = ""
+response = completion(
+    model="fireworks_ai/mixtral-8x7b-instruct", 
+    messages=[
+       {"role": "user", "content": "hello from litellm"}
+   ],
+    stream=True
+)
+
+for chunk in response:
+    print(chunk)
+```
+
+
+## Supported Models - ALL Fireworks AI Models Supported!
+We support ALL Fireworks AI models, just set `fireworks_ai/` as a prefix when sending completion requests
+
+| Model Name               | Function Call                                                                                                                                                      |
+|--------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------|
+| mixtral-8x7b-instruct | `completion(model="fireworks_ai/mixtral-8x7b-instruct", messages)` | 
+| firefunction-v1 | `completion(model="fireworks_ai/firefunction-v1", messages)` |
+| llama-v2-70b-chat | `completion(model="fireworks_ai/llama-v2-70b-chat", messages)` |  

docs/my-website/sidebars.js

@@ -138,6 +138,7 @@ const sidebars = {
         "providers/ollama", 
         "providers/perplexity", 
         "providers/groq", 
+        "providers/fireworks_ai", 
         "providers/vllm", 
         "providers/xinference", 
         "providers/cloudflare_workers", 

litellm/__init__.py

@@ -328,6 +328,7 @@ openai_compatible_providers: List = [
     "perplexity",
     "xinference",
     "together_ai",
+    "fireworks_ai",
 ]
 
 
@@ -479,6 +480,7 @@ provider_list: List = [
     "voyage",
     "cloudflare",
     "xinference",
+    "fireworks_ai",
     "custom",  # custom apis
 ]
 

litellm/main.py

@@ -892,6 +892,7 @@ def completion(
             or custom_llm_provider == "mistral"
             or custom_llm_provider == "openai"
             or custom_llm_provider == "together_ai"
+            or custom_llm_provider in litellm.openai_compatible_providers
             or "ft:gpt-3.5-turbo" in model  # finetune gpt-3.5-turbo
         ):  # allow user to make an openai call with a custom base
             # note: if a user sets a custom base - we should ensure this works
@@ -2394,6 +2395,7 @@ async def aembedding(*args, **kwargs):
             or custom_llm_provider == "deepinfra"
             or custom_llm_provider == "perplexity"
             or custom_llm_provider == "groq"
+            or custom_llm_provider == "fireworks_ai"
             or custom_llm_provider == "ollama"
             or custom_llm_provider == "vertex_ai"
         ):  # currently implemented aiohttp calls for just azure and openai, soon all.
@@ -2893,6 +2895,7 @@ async def atext_completion(*args, **kwargs):
             or custom_llm_provider == "deepinfra"
             or custom_llm_provider == "perplexity"
             or custom_llm_provider == "groq"
+            or custom_llm_provider == "fireworks_ai"
             or custom_llm_provider == "text-completion-openai"
             or custom_llm_provider == "huggingface"
             or custom_llm_provider == "ollama"

litellm/tests/test_completion.py

@@ -529,6 +529,25 @@ def test_completion_azure_gpt4_vision():
 # test_completion_azure_gpt4_vision()
 
 
+def test_completion_fireworks_ai():
+    try:
+        litellm.set_verbose = True
+        messages = [
+            {"role": "system", "content": "You're a good bot"},
+            {
+                "role": "user",
+                "content": "Hey",
+            },
+        ]
+        response = completion(
+            model="fireworks_ai/accounts/fireworks/models/mixtral-8x7b-instruct",
+            messages=messages,
+        )
+        print(response)
+    except Exception as e:
+        pytest.fail(f"Error occurred: {e}")
+
+
 @pytest.mark.skip(reason="this test is flaky")
 def test_completion_perplexity_api():
     try:

litellm/utils.py

@@ -5375,6 +5375,17 @@ def get_llm_provider(
                 # groq is openai compatible, we just need to set this to custom_openai and have the api_base be https://api.groq.com/openai/v1
                 api_base = "https://api.groq.com/openai/v1"
                 dynamic_api_key = get_secret("GROQ_API_KEY")
+            elif custom_llm_provider == "fireworks_ai":
+                # fireworks is openai compatible, we just need to set this to custom_openai and have the api_base be https://api.groq.com/openai/v1
+                if not model.startswith("accounts/fireworks/models"):
+                    model = f"accounts/fireworks/models/{model}"
+                api_base = "https://api.fireworks.ai/inference/v1"
+                dynamic_api_key = (
+                    get_secret("FIREWORKS_API_KEY")
+                    or get_secret("FIREWORKS_AI_API_KEY")
+                    or get_secret("FIREWORKSAI_API_KEY")
+                    or get_secret("FIREWORKS_AI_TOKEN")
+                )
             elif custom_llm_provider == "mistral":
                 # mistral is openai compatible, we just need to set this to custom_openai and have the api_base be https://api.mistral.ai
                 api_base = (