Inline code completion and OpenAI-compatible chat completion endpoints for drop-in integration with editors and tooling.

This is the MisarCoder gateway

These endpoints are served by the MisarCoder MoE gateway at https://api.misar.io/coder, not by the MisarDev product API at https://api.misar.io/dev. See the MisarCoder docs for the full gateway reference.

Inline Completions

POST/complete

Generate a code completion given a cursor position represented by a prefix and suffix. Requires Authorization: Bearer YOUR_API_KEY.

Request body

prefixstringbodyrequired

All text before the cursor.

suffixstringbodyrequired

All text after the cursor.

languagestringbodyrequired

Language identifier (e.g. typescript, python, go).

file_pathstringbody

Relative path of the file — used for additional context.

Response fields

completionstring

The text to insert at the cursor position.

finish_reasonstring

"stop" — model finished naturally; "length" — truncated at token limit.

Request

{
  "prefix": "def calculate_total(items):\n    total = 0\n    for item in items:\n        total += ",
  "suffix": "\n    return total",
  "language": "python",
  "file_path": "src/utils.py"
}

200 — OK

{
  "completion": "item.price * item.quantity",
  "finish_reason": "stop"
}

Inline completions are optimised for low latency. The model uses a fill-in-the-middle (FIM) prompt format internally — you only need to supply the raw prefix and suffix.

OpenAI-Compatible Chat Completions

POST/v1/chat/completions

A drop-in compatible endpoint for tools and libraries that target the OpenAI Chat Completions API. Requires Authorization: Bearer YOUR_API_KEY.

Request body

modelstringbodyrequired

Model identifier — misar-flash (fast), misar-pro (quality), or online (web-grounded synthesis via misar-flash). See Models.

messagesArray<{role, content}>bodyrequired

Conversation messages with role (system, user, assistant) and content.

max_tokensnumberbody

Maximum tokens in the model response.

temperaturenumberbody

Sampling temperature.

streambooleanbody

When true, returns Server-Sent Events in the OpenAI delta format.

Response fields

idstring

Completion identifier, e.g. chatcmpl-abc123.

objectstring

chat.completion for non-streaming, chat.completion.chunk for streamed deltas.

creatednumber

Unix timestamp of creation.

modelstring

The model that produced the response.

choicesArray<object>

Completion choices. Each contains index, message (or delta when streaming), and finish_reason.

usageobject

Token counts: prompt_tokens, completion_tokens, total_tokens.

Request

{
  "model": "misar-flash",
  "messages": [
    { "role": "system", "content": "You are a helpful coding assistant." },
    { "role": "user", "content": "Explain what a closure is in JavaScript." }
  ],
  "max_tokens": 1024,
  "temperature": 0.2,
  "stream": true
}

200 — OK (non-streaming)

{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "created": 1740000000,
  "model": "misar-flash",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "A closure is a function that retains access to..."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 42,
    "completion_tokens": 198,
    "total_tokens": 240
  }
}

Streaming

Set "stream": true to receive Server-Sent Events in the standard OpenAI delta format:

data: {"id":"chatcmpl-abc123","object":"chat.completion.chunk","choices":[{"delta":{"content":"A closure"},"index":0}]}

data: {"id":"chatcmpl-abc123","object":"chat.completion.chunk","choices":[{"delta":{"content":" is a"},"index":0}]}

data: [DONE]

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="https://api.misar.io/coder/v1"
)

response = client.chat.completions.create(
    model="misar-flash",
    messages=[{"role": "user", "content": "Write a binary search in Python"}],
    stream=True
)

for chunk in response:
    print(chunk.choices[0].delta.content or "", end="")

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "YOUR_API_KEY",
  baseURL: "https://api.misar.io/coder/v1",
});

const stream = await client.chat.completions.create({
  model: "misar-flash",
  messages: [{ role: "user", content: "Write a binary search in TypeScript" }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

curl https://api.misar.io/coder/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "misar-flash",
    "messages": [{"role": "user", "content": "Write a binary search in Python"}],
    "stream": true
  }'

Use base_url="https://api.misar.io/coder/v1" with any OpenAI-compatible SDK. The endpoint accepts the same request shape and returns the same response format.

Models

misar-flash and misar-pro are Misar's own self-hosted models. The same two models are also published in the public assisters.dev catalogue.

Model	Speed	Best for
`misar-flash`	Fastest	Everyday chat, inline assistance, and low-latency tasks.
`misar-pro`	Higher quality	Reasoning, longer answers, and quality-sensitive generation.
`online`	—	Web-grounded synthesis: retrieves live web results, then answers with `misar-flash`. For structured answers with source citations, use the dedicated Search API instead.

Text embeddings are produced by a local 768-dimension embedding model. It powers workspace RAG and is selected automatically — you do not pass it as a model.

Status codes

Status	Meaning
`200 OK`	Completion returned (or the SSE stream opened when `stream: true`).
`400 Bad Request`	Malformed body, unknown `model`, or missing `messages`.
`401 Unauthorized`	Missing, malformed, revoked, or wrong-product API key.
`429 Too Many Requests`	Rate limit exceeded — retry after the `Retry-After` header.

See Authentication for the product-binding rules that make a non-MisarDev key return 401.

Completions API

Inline Completions

OpenAI-Compatible Chat Completions

Streaming

Models

Status codes

On this page