Millisa / Interactive Chat
Millisa Chat
Production
Active Model:

Developer Authentication Required

Authenticate with your developer credentials to issue custom scoped Millisa API keys, configure IP whitelists, and monitor your monthly quota.

Developer Integration Hub

Interactive OpenAPI 3.1 Swagger reference, real-time endpoint testing, and multi-platform SDKs.

Full-Screen Docs
⚡ 100% OpenAI & Gemini Drop-In Compatible

Integrate Millisa into Any Project in 30 Seconds

Millisa provides a high-throughput, enterprise-grade OpenAI-compatible REST API gateway. You can use official SDKs (openai in Python, Node.js, Go, PHP, etc.) or native HTTP clients by simply pointing your Base URL to https://millisa.codly.in/api/v1.

1

Get Your API Key

Generate a secret bearer key in the Keys & Quotas dashboard tab.

2

Set Base URL

https://millisa.codly.in/api/v1

3

Select Engine Model

Pass millisa-gemini-flash, millisa-gpt-4o, or millisa-auto.

Developer Sign-In Required

Sign in to view real-time token telemetry, prompt versus completion volume, model consumption distributions, and request debit logs.

Developer Sign-In Required

Authenticate with your developer account to access your Millisa Token Wallet, claim your 10 free welcome tokens, and manage pay-as-you-go balance across Gemini, Claude, GPT, and Grok.

Wallets & Revenue Governance

Super Admin Master Console Codly Gateway Active

Real-time revenue telemetry, Codly payment gateway transactions, token minting ledger, and developer wallet balance controls.

Total Sales Revenue
₹0.00
0 paid enterprise orders Codly Live
₹30,000 Growth Tier
0
₹0 total volume 60,000 MT / pack
₹50,000 Scale Tier (Bonus)
0
₹0 total volume +25k Bonus
MT in Circulation
0.00 MT
Active unspent wallet volume Reserve Healthy

Customer Purchases & Gateway Settlements

Codly.in Live Sync
0 total orders recorded
Order / Reference ID Customer / Developer Plan Purchased Price (INR) Tokens Credited Status Actions
Loading purchase orders...

Developer Accounts & Token Balances

Master Ledger
0 accounts registered
Developer Role / Tier Available Balance Lifetime Funded Lifetime Consumed Status Actions
Loading developer wallets...

Inference Command Center

Real-time upstream provider connectivity, latency benchmarks, and live request stream.

Google Gemini API

Primary Inference Provider
Operational

Endpoints online and responding normally.

Round-Trip Latency: ⚡ 177 ms

OpenAI Platform

Multimodal & Flagship Reasoning
Not Configured

Master OpenAI key pending configuration in vault.

Round-Trip Latency: N/A

Anthropic Claude

Advanced Code & Synthesis
Not Configured

Master Claude key pending configuration in vault.

Round-Trip Latency: N/A

xAI Grok Platform

Real-Time Intelligence & Frontier
Not Configured

Master Grok key pending configuration in vault.

Round-Trip Latency: N/A

Model Gateway Routing Matrix

Global models served across completions and agent query endpoints.
Model Identifier Upstream Provider Context Window Speed Tier Failover Strategy Status
millisa-auto Smart Multi-Engine 1,000,000 tokens Ultra Fast Primary: Gemini Flash · Fallback: GPT-4o-mini Active
millisa-gemini-flash Google Gemini 1,000,000 tokens Ultra Fast Direct Google Gemini 1.5 Flash Active
millisa-gemini-pro Google Gemini 2,000,000 tokens Reasoning Direct Google Gemini 1.5 Pro Active
millisa-gpt-4o-mini OpenAI Platform 128,000 tokens Ultra Fast Direct OpenAI GPT-4o Mini Active
millisa-gpt-4o OpenAI Platform 128,000 tokens Flagship Direct OpenAI GPT-4o Flagship Active
millisa-claude-3-5-sonnet Anthropic Claude 200,000 tokens Advanced Direct Claude 3.5 Sonnet Active
millisa-claude-3-haiku Anthropic Claude 200,000 tokens Ultra Fast Direct Claude 3 Haiku Active
millisa-grok-2 xAI Grok 128,000 tokens Reasoning Direct xAI Grok 2 Active
millisa-grok-beta xAI Grok 128,000 tokens Ultra Fast Direct xAI Grok Beta Active

Live Request Telemetry

Real-time stream of incoming API completions with latency and token metrics.
Status Model & Engine Latency Tokens Client IP Timestamp Inspection

Developer Accounts & Visitor Telemetry

Oversee registered developers, audit real-time model usage with timestamps, and monitor guest web visitor geolocation.

User Group & Tier Usage & Cost Models Used Assigned Models Last Active Status Actions

Master Key Vault & Engine Configuration

Configure upstream provider keys and system prompts. Master keys remain strictly on the server.

Upstream Provider Vault

Configure master keys for multi-model inference. Click "Test Connection" to test real-time connectivity against each upstream API.
Google Gemini
Gemini 1.5 Flash, 1.5 Pro, 2.0 Flash · Primary Inference
OpenAI Platform
GPT-4o, GPT-4o Mini, Embeddings · Multimodal Flagship
Anthropic Claude
Claude 3.5 Sonnet, Claude 3 Haiku · Deep Code & Reasoning
xAI Grok Platform
Grok 2, Grok Beta · Real-Time Intelligence & Frontier Reasoning
Groq Cloud LPU
LLaMA 3.3 70B, LLaMA 3.1 8B, Mixtral 8x7B · Ultra-Low Latency Inference
Mistral AI Platform
Mistral Large, Mistral Small, Codestral, Pixtral · Sovereign European AI
Cloudflare Workers AI
Edge-Native Serverless AI · LLaMA 3.3 70B, LLaMA 3.1 8B, Mistral 7B
Hugging Face Hub & Serverless Router
Qwen 2.5 72B, LLaMA 3.2, DeepSeek · Open Weights Ecosystem
NVIDIA API & NIM Acceleration
Nemotron 70B, LLaMA 3.3 70B NIM, Mistral Large 2 · TensorRT Accelerated
Razorpay Gateway & Millisa Wallet Billing
Accept UPI, Credit/Debit Cards, NetBanking from Developers (INR)
Determines default dispatch for requests targeting millisa-auto.

Resume Studio

Logged-in users build ATS-friendly LaTeX resumes and download the PDF. Each new resume costs a fixed number of Millisa Tokens, charged only after the PDF is created. Recompiling and downloads are free. Super Admin is never charged. Documents generated from a Training Studio agent's master template (format kept, content written fresh section by section) have their own price, charged once per document; rewrites and re-downloads of that document are free.

Documents Converter

Users convert files in the chat (attach a file and ask, or + → Convert file). PDF → Word (with or without OCR), searchable PDF, PDF → Excel/JPG/text, images → PDF/Word, merge, split, rotate, page numbers and watermark run in the user's browser (free, files never leave the device). Word, PowerPoint and Excel → PDF, compress, protect and unlock use iLovePDF (logged-in users; files go straight from the browser to iLovePDF). Best-quality PDF → Word, PowerPoint and Excel use Adobe PDF Services (500 free a month; keys at developer.adobe.com/document-services). iLovePDF keys: developer.ilovepdf.com → Projects.

Study Assistant Extension · Developer mode

For testing only. While on, the extension connected to the accounts you tick below also works on live quiz games (Kahoot, Quizizz, quiz.com …), Moodle / Canvas quizzes, graded Google Forms and timed quiz pages; the panel shows a DEV MODE badge. Every other user stays blocked as usual. Exam and proctoring sites stay blocked even here. Each answer on a normally blocked page is listed below. Test on games and quizzes you host yourself.
Accounts with developer mode (their extension is connected):
Loading…

Code Runner

Users write and run code inside the chat. Python, JavaScript, TypeScript, C, C++, SQL, PHP, Ruby, R and HTML/CSS run in the user's own browser (free, nothing reaches this server; C/C++ download a 105 MB compiler on first use). Java, C#, Go, Rust, Kotlin, Swift, Scala and Dart run on a Piston code runner on a separate server (logged-in users, 10 s and 256 MB per run, no internet). Running code is free; Millisa's help on the code is charged as normal chat.

Word Documents

Logged-in users ask Millisa for a finished Word (.docx) document (page borders and numbers from any page, fonts, spacing, lists, tables, table of contents) or edit an uploaded one. A fixed price is charged once per document when it is downloaded; re-downloads and revisions of the same file in the same chat are free. Super Admin is never charged. Documents generated from a Training Studio agent's master template (format kept, content written fresh section by section) have their own price, charged once per document; rewrites and re-downloads of that document are free.

Default Models per User Group

Which image and text model each group gets. If the chosen image model fails, free FLUX / Stable Diffusion models are used as backup. Smart Auto picks the cheapest text model that can answer each question. You can override any single user or guest in the Developer Accounts tab.
User group Image generation model Text response model
Loading…
Paid user = has completed at least one token purchase, or has the Pro / Enterprise tier. Guests are identified by IP address + device. Developer API requests use the API key owner's models.

Agent Persona & System Prompt

Injected into all completion requests as the system role instruction.

Platform Governance Defaults