Reka

Overview

Property	Details
Description	Reka serves its own models and a curated selection of open models over one OpenAI-compatible API, with automatic prompt caching and no platform fee or markup.
Provider Route on LiteLLM	reka/
Link to Provider Doc	Reka Developer Reference ↗
Base URL	https://api.reka.ai/v1
Supported Operations	/chat/completions, /responses, /messages

We support ALL Reka models, just set reka/ as a prefix when sending requests

Available Models

Model	Context	Max output
reka/reka-flash-3	64k	58,982
reka/reka-edge-2603	16k	14,745
reka/deepseek4-flash	1M	384,000
reka/deepseek-v4-pro	1M	393,216
reka/glm5.3	262k	131,072
reka/glm5.3-flash	262k	131,072
reka/qwen3.8-27b	262k	131,072

Required Variables

Environment Variables

os.environ["REKA_API_KEY"] = ""  # your Reka API key
os.environ["REKA_API_BASE"] = ""  # optional, defaults to https://api.reka.ai/v1

Usage - LiteLLM Python SDK

Non-streaming

Reka Non-streaming Completion

import os
from litellm import completion

os.environ["REKA_API_KEY"] = ""  # your Reka API key

response = completion(
    model="reka/reka-flash-3",
    messages=[{"role": "user", "content": "Hello, how are you?"}],
)

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

Streaming

Reka Streaming Completion

import os
from litellm import completion

os.environ["REKA_API_KEY"] = ""  # your Reka API key

response = completion(
    model="reka/reka-flash-3",
    messages=[{"role": "user", "content": "Write a short story about AI"}],
    stream=True,
)

for chunk in response:
    print(chunk)

Vision

reka-edge-2603 accepts images and video. Image input uses the OpenAI content-part shape; video uses a video_url part the same way. Check input_modalities on GET /v1/models before sending media to any other model.

Reka Image Input

import os
from litellm import completion

os.environ["REKA_API_KEY"] = ""  # your Reka API key

response = completion(
    model="reka/reka-edge-2603",
    messages=[{
        "role": "user",
        "content": [
            {"type": "text", "text": "What animal is this? Answer briefly."},
            {"type": "image_url", "image_url": {"url": "https://v0.docs.reka.ai/_images/000000245576.jpg"}},
        ],
    }],
)

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

Responses API

Reka does not serve /v1/responses natively, so LiteLLM translates litellm.responses calls into Reka chat completions and converts the result back into a Responses API object.

Reka Responses API

import os
import litellm

os.environ["REKA_API_KEY"] = ""  # your Reka API key

response = litellm.responses(
    model="reka/reka-flash-3",
    input="Say hello",
)

print(response.output_text)

Usage - LiteLLM Proxy Server

config.yaml

model_list:
  - model_name: reka-flash-3
    litellm_params:
      model: reka/reka-flash-3
      api_key: os.environ/REKA_API_KEY
  - model_name: reka-edge
    litellm_params:
      model: reka/reka-edge-2603
      api_key: os.environ/REKA_API_KEY

general_settings:
  master_key: os.environ/LITELLM_MASTER_KEY

Start the proxy:

Start LiteLLM Proxy

export REKA_API_KEY="your-api-key"
export LITELLM_MASTER_KEY="sk-local-reka"
litellm --config config.yaml --port 4000

# RUNNING on http://0.0.0.0:4000

A deployment configured this way serves /v1/chat/completions, /v1/responses, and /v1/messages on the proxy.

Chat Completions

curl

curl http://localhost:4000/v1/chat/completions \
  -H "Authorization: Bearer $LITELLM_MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "reka-flash-3",
    "messages": [{"role": "user", "content": "Hello, how are you?"}]
  }'

OpenAI SDK

Reka via Proxy - OpenAI SDK

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:4000",
    api_key="sk-local-reka",
)

response = client.chat.completions.create(
    model="reka-flash-3",
    messages=[{"role": "user", "content": "hello from litellm"}],
)

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

Responses

curl

curl http://localhost:4000/v1/responses \
  -H "Authorization: Bearer $LITELLM_MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "reka-flash-3",
    "input": "Hello, how are you?"
  }'

Messages

curl

curl http://localhost:4000/v1/messages \
  -H "x-api-key: $LITELLM_MASTER_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "reka-edge",
    "max_tokens": 64,
    "messages": [{"role": "user", "content": "Hello, how are you?"}]
  }'

You can also add Reka from the Admin UI. Go to Models, then Add Model, pick Reka as the provider, enter a reka/ model id, and paste your key.

Cost Tracking

Reka models are not yet in LiteLLM's model cost map, so spend is not computed automatically. Reka bills per token at each model's rate with no platform fee, and publishes the rates on developer.reka.ai/models and in the pricing object of GET /v1/models (US dollars per token, as strings: prompt, completion, and input_cache_read). Pass those values as input_cost_per_token and output_cost_per_token on the deployment and LiteLLM will track spend for it, returning the amount in the x-litellm-response-cost response header and recording it in spend logs. Reka lists rates per million tokens; divide by 1,000,000 for the per-token value.

config.yaml

model_list:
  - model_name: reka-edge
    litellm_params:
      model: reka/reka-edge-2603
      api_key: os.environ/REKA_API_KEY
      input_cost_per_token: 0.0000001   # $0.10 / 1M
      output_cost_per_token: 0.0000001  # $0.10 / 1M
  - model_name: glm5.3-flash
    litellm_params:
      model: reka/glm5.3-flash
      api_key: os.environ/REKA_API_KEY
      input_cost_per_token: 0.00000015  # $0.15 / 1M
      output_cost_per_token: 0.0000005  # $0.50 / 1M

Where a model supports prompt caching, Reka caches repeated prompt prefixes automatically and bills those tokens at the cached-input rate; usage.reasoning_tokens is included inside completion_tokens and is not billed twice.

Custom API Base

Option 1: Environment variable

Custom API Base via env var

import os
from litellm import completion

os.environ["REKA_API_BASE"] = "https://custom.reka.example/v1"
os.environ["REKA_API_KEY"] = ""  # your API key

response = completion(
    model="reka/reka-flash-3",
    messages=[{"role": "user", "content": "Hello!"}],
)

Option 2: Pass directly

Custom API Base via parameter

from litellm import completion

response = completion(
    model="reka/reka-flash-3",
    messages=[{"role": "user", "content": "Hello!"}],
    api_base="https://custom.reka.example/v1",
    api_key="your-api-key",
)

Passing api_base="https://api.reka.ai/v1" without the reka/ prefix also resolves to the Reka provider, so model="reka-flash-3" with that base URL is routed as reka and picks up REKA_API_KEY.