One key. Four live models.

Use two chat models, embeddings and speech transcription through one OpenAI-compatible endpoint. Every example below tracks the catalog available today.

Get your API key Make your first call

4 live models

Base URLhttps://bharatcode.ai/api/model/v1

Bearer authentication · OpenAI-compatible payloads

Start with the catalog

Ask the API what is live before choosing a model. The response is authoritative; retired or unavailable services are not advertised.

Export your keyKeep it in a server-side environment variable.
GET /modelsRead the current model IDs and capabilities.
Send an exact model IDEvery generation request requires the model field.

$ export BHARATCODE_API_KEY=bc_live_your_key_here
$ curl https://bharatcode.ai/api/model/v1/models -H "Authorization: Bearer $BHARATCODE_API_KEY"

Models available now

The public surface is deliberately small. Choose the model by the job, then use its matching endpoint. Access requirements are shown alongside each model.

Live catalog

ModelUse it forEndpointAccess

GLM 5.3z-ai/glm-5.3

Long-context reasoning and tool use/chat/completionsPaid subscribers

BharatCode Qwen3.6 35Bbharatcode:qwen36-35b-awq-200k

Coding, reasoning and tool use/chat/completionsEligible accounts

BharatCode Embed Smallbharatcode:embed-small-v1

Semantic search and retrieval/embeddingsEligible accounts

Parakeet TDT 0.6B v3parakeet-tdt-0.6b-v3

Multilingual speech to text/audio/transcriptionsEligible accounts

Copy a working request

Each example includes the required model field. Expand only the endpoint you need.

POSTGenerate code and call tools/chat/completions

Send OpenAI-compatible messages to Qwen 3.6 or GLM 5.3. The model field is required; GLM 5.3 requires a paid subscription.

$ curl https://bharatcode.ai/api/model/v1/chat/completions -H "Authorization: Bearer $BHARATCODE_API_KEY" -H "Content-Type: application/json" -d '{"model":"bharatcode:qwen36-35b-awq-200k","messages":[{"role":"user","content":"Say hello from BharatCode"}]}'
$ curl https://bharatcode.ai/api/model/v1/chat/completions -H "Authorization: Bearer $BHARATCODE_API_KEY" -H "Content-Type: application/json" -d '{"model":"z-ai/glm-5.3","messages":[{"role":"user","content":"Design a migration plan for this service"}]}'

POSTCreate text embeddings/embeddings

Turn text into vectors for semantic search, retrieval and project matching.

$ curl https://bharatcode.ai/api/model/v1/embeddings -H "Authorization: Bearer $BHARATCODE_API_KEY" -H "Content-Type: application/json" -d '{"model":"bharatcode:embed-small-v1","input":"student project search query"}'

POSTTranscribe an audio file/audio/transcriptions

In Postman, choose Body → form-data. Add model as Text and file as File; do not send this request as JSON.

$ curl https://bharatcode.ai/api/model/v1/audio/transcriptions -H "Authorization: Bearer $BHARATCODE_API_KEY" -F model=parakeet-tdt-0.6b-v3 -F file=@sample.wav

Use the OpenAI SDK

Change only the base URL and model. Keep your BharatCode key in the backend environment.

$ import OpenAI from 'openai'
$ 
$ const client = new OpenAI({
$   apiKey: process.env.BHARATCODE_API_KEY,
$   baseURL: 'https://bharatcode.ai/api/model/v1',
$ })
$ 
$ const completion = await client.chat.completions.create({
$   model: 'bharatcode:qwen36-35b-awq-200k',
$   messages: [{ role: 'user', content: 'Build a study planner schema' }],
$ })

Hand this to your coding agent

Current routes, models and security rules in one paste-ready block.

- Use BharatCode as an OpenAI-compatible model endpoint.
- Get the user's developer API key from https://bharatcode.ai/account.
- Set baseURL to https://bharatcode.ai/api/model/v1.
- Send Authorization: Bearer <BHARATCODE_API_KEY>.
- Use bharatcode:qwen36-35b-awq-200k for coding and tool-use tasks.
- Paid subscribers can use z-ai/glm-5.3 for long-context reasoning and tool-use tasks.
- Use bharatcode:embed-small-v1 at /embeddings for text embeddings.
- Use parakeet-tdt-0.6b-v3 at /audio/transcriptions for speech-to-text.
- For transcription, send multipart form-data fields named model and file.
- Keep keys on the server side. Do not ship them in browser bundles or mobile apps.

Shared compute, used carefully

The API shares BharatCode serving capacity with Chat, CLI, Desktop and VS Code. Avoid unattended request loops, handle rate and availability errors, and treat your key like a password. GLM 5.3 uses a shared one-request queue; if capacity is temporarily unavailable, retry a 503 Model unavailable. later.