Tesseracted Labs / tesseracted/kolibri-1
Kolibri-1 - access through LLMTR
Kolibri-1 is an open-weight (Apache 2.0) model of European origin, built for German and English by Germany-based Aleph Alpha; Turkish is not among its supported languages. Input and output together must fit in 32,768 tokens; one response returns at most 16,384 tokens and is capped at 4,096 when max_tokens is omitted. The conversation history is limited to 80 messages and 60,000 characters. It accepts text only; JSON mode and JSON Schema output are not supported. Tool calling works; for a turn that must not call a tool, leave out the tools field instead of sending tool_choice none. Reasoning is off by default; setting reasoning_effort to low, medium or high turns it on, and reasoning tokens count toward the output limit. The free period ends on 9 October 2026 at 23:59 (TRT); access to the model closes after that.
Technical specifications
| Canonical ID | tesseracted/kolibri-1 |
|---|---|
| Provider | Tesseracted Labs |
| Context window | 32,768 tokens |
| Operations | CHAT_COMPLETIONS |
| Modalities | text |
Pricing
An 8% platform margin applies to credit top-ups; model usage prices are not separately marked up.
| Operation | Metric | Unit | Price |
|---|---|---|---|
| CHAT_COMPLETIONS | INPUT_TEXT | PER_1M_TOKENS | Not available |
| CHAT_COMPLETIONS | OUTPUT_TEXT | PER_1M_TOKENS | Not available |
Example usage
With existing OpenAI SDK flows, change only the base URL and model identifier.
curl https://llmtr.com/v1/chat/completions -H "Authorization: Bearer llmtr-your_key" -H "Content-Type: application/json" -d '{"model":"tesseracted/kolibri-1","messages":[{"role":"user","content":"Hello"}]}'Guides about this model
- Kolibri-1 API: free on LLMTR until 9 October, first request and limits - Send an OpenAI-compatible Chat Completions request with tesseracted/kolibri-1. The model is free and needs no top-up; input and output together must fit in 32,768 tokens.
- Kolibri-1 tool calling and reasoning_effort: a developer guide for the LLMTR API - Reasoning is off by default on Kolibri-1; setting reasoning_effort to low, medium or high turns it on and it counts toward the output limit. Tool calling works; every call in the history needs its result.
- What is Aleph Alpha Kolibri? An open-weight model trained in Germany - Kolibri is an MoE model with 78.1 billion parameters, 3.46 billion of which run per token. It was trained for German and English, and its weights are open under Apache 2.0.
- Running Kolibri on your own servers: hardware, vLLM and context settings - Kolibri needs about 78 GB of memory with FP8 weights; two H100s or a single H200 is the minimum hardware. It is served through vLLM with Aleph Alpha's plugin.
- Does Kolibri support Turkish? Sovereign open models for teams in Türkiye - Kolibri officially supports German and English; Turkish is not on that list. Open weights let you choose where the model runs, but Turkish quality has to be measured separately.
Related models
- Gemma 4 llmtr/gemma-4
- Qwen 3.6 35B-A3B llmtr/qwen3-6-35b
- Trendyol Asure 12B llmtr/trendyol-asure-12b
- Magibu 11B v8 llmtr/magibu-11b-v8
- Muse Glimmer 30B (Turkey) llmtr/muse-glimmer-30b-tr
- Leyla llmtr/leyla
- EmbeddingGemma 300M llmtr/embeddinggemma-300m
- Qwen 3.5 4B llmtr/qwen3-5-4b