Plannix 3.6
Plannix 3.6 is Plannix's frontier model built for coding, agentic tasks, and knowledge work.
Using the API
If you already have an API key, set the model name to plannix-3.6-ultra:
curl https://plannix.web.id/api/v1/responses \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $PLANNIX_API_KEY" \
-d '{
"model": "plannix-3.6-ultra",
"input": "Find and fix the bug, then explain it: function median(a) { a.sort(); return a[a.length / 2]; }"
}'New to the Plannix API? Follow the Quickstart to create an account and make your first request.
At a glance
| Property | Value |
|---|---|
| Model name | plannix-3.6-ultra |
| Context window | 1,000,000 tokens |
| Knowledge cutoff | October 2026 |
| Modalities | Text and structured AST input; text output |
| Output limit | 128,000 tokens per request |
| Input price | $0.80 / 1M tokens |
| Output price | $2.40 / 1M tokens |
| Reasoning | Low, medium, high (default), or xhigh |
| APIs | Responses API, Chat Completions |
| Tools | Function calling, web search, code execution, database & ERD modeling |
Rate limits and live pricing for your team are on the Model detail page and Pricing.
Important details
We highly recommend setting a prompt_cache_key (Responses API; x-plannix-conv-id header on Chat Completions). It routes a conversation's requests to the same server node on our Hostinger cloud infrastructure, making cache hits reliable; without it you often pay full input price on a cache-cold server. See Multi-turn Best Practices for common patterns.
Long agent loops additionally benefit from context compaction; for tool-heavy workloads see function calling.
Reasoning chain-of-thought is always preserved on the Responses API. Responses from plannix-3.6-ultra include reasoning.thought_tokens so multi-turn conversations keep the model's reasoning without extra configuration. Chat Completions is unchanged.
Fast variant
Plannix 3.6 Instant is the ultra-low latency variant served on high-throughput Groq LPU™ clusters. It is optimized for live autocomplete, real-time typing in IDEs, and sub-100ms API webhooks. Available directly in Plannix Build and API console. Rates are on the Pricing page.
Where it runs
- Plannix API: get a key from the Developer Console
- Regional endpoint: served at
https://plannix.web.id/api/v1with contractual zero data retention guarantee - Plannix Build: the default model of the autonomous coding agent
- IDE Extensions: available in VS Code, Cursor, and Windsurf via OpenAI-compatible endpoints
- Model gateways: OpenRouter, Vercel AI SDK, LangChain, and LlamaIndex