Kolibri-1
Aleph Alpha’s open-weight English–German MoE. Serve with vLLM plus the aleph-alpha-inference package (reasoning and tool-calling parsers enabled) for an OpenAI-compatible endpoint.
PricePrice not published
Checked3 October 2026
Knows up to June 2026
1,923events after its cutoff
84 historic, 451 major. 142 days of news.
It launched on 3 October 2026, already missing 664 of them.
How to call it
2 ways to use Kolibri-1.
| Provider | Model id | Endpoint | Docs |
|---|---|---|---|
| Hugging Face | huggingface.co | ||
| Self-hosted (vLLM + aleph-alpha-inference, OpenAI-compatible API) | huggingface.co |
Pricing
Price not published.
No price source linked. Checked 3 October 2026.
Specs
- Input
- Text
- Output
- Text
- Context window
- 1,048,576 tokens
- Max output
- Not published
- Open weights
- Yes
- Licence
- Apache-2.0
- Training cutoff
- June 2026
- Released
- 3 October 2026
- Status
- Current
- Type
- Reasoning model
What stands out
What its maker marketed at launch, and what people found later, each with its source.
German-optimised tokenizer
Bilingual EN–DE 128k tokenizer; Aleph Alpha reports 11.2% fewer German tokens than GPT-5’s tokenizer (an independent test on the German Basic Law found ~15%).
Sparse 1M-token context
78B-total / ~3.46B-active MoE (384 experts, 6 routed) with 512-token sliding-window attention and a global layer every 5th layer; up to 1,048,576 tokens of context (262,144 recommended).
Abstention / grounding training
Trained with a Merlin–Arthur protocol to abstain rather than hallucinate; Aleph Alpha reports an M/A grounding score of 0.23 (scale 0–0.5).
Four assistants, one scale
Kolibri-1 is added as a fifth row. Each row runs from what the model knows (graphite) through the months between its cutoff and its launch (hatched) to today (pink).
- GPT-6.2 Astra 1,915 at launch, 1,958 today
- Claude Opus 5.5 372 at launch, 1,923 today
- Gemini 4 Hydrogen 1,880 at launch, 2,018 today
- Grok 4.7 388 at launch, 1,958 today
- Kolibri-1 664 at launch, 1,923 today
Show as a table
| Model | Cutoff | Released | Missing at launch | Missing today |
|---|---|---|---|---|
| GPT-6.2 Astra | May 2026 | 17 November 2026 | 1,915 | 1,958 |
| Claude Opus 5.5 | June 2026 | 22 September 2026 | 372 | 1,923 |
| Gemini 4 Hydrogen | March 2026 | 12 November 2026 | 1,880 | 2,018 |
| Grok 4.7 | May 2026 | 21 September 2026 | 388 | 1,958 |
| Kolibri-1 | June 2026 | 3 October 2026 | 664 | 1,923 |
Notes
Reasoning mode with configurable effort and tool calling. Recommended sampling temperature 1.0, top_p 0.97, top_k 128. Needs ~2×80GB GPUs (min 2× A100 80GB). Benchmarks are Aleph Alpha’s own (AIME 2025 96.9 EN, GPQA Diamond 84.3 EN). Weaker than Qwen models on closed-book recall and agentic coding per reviewers. No hosted API price found.