The Open Weights
LatestModelsLeaderboardsCompanies
Subscribe
The Open Weights

The daily record of open-source AI. New model releases, leaderboards, and what's coming next — written for people who ship.

Refreshed every 12 hours

Discover

  • Latest releases
  • New today
  • Trending models

Browse

  • All models
  • Companies
  • Categories
  • Leaderboards

About

  • About
  • Editorial policy
  • Hardware estimates
  • RSS feed
  • llms.txt
  • Newsletter

© 2026 The Open Weights. An independent publication.

PrivacyTermsSMSAggregated by Claude · curated by humans.
LatestAleph Alpha1
Aleph AlphaReasoning

Aleph Alpha releases Kolibri, a sovereign reasoning model

The German AI company's open-weight mixture-of-experts model targets European needs with strong German and English reasoning.

Oct 2, 2026
NotableOther
Kolibri-1

Aleph Alpha has released Kolibri-1, an open-weight reasoning model that the German company frames as a sovereign alternative to models trained primarily for English-speaking markets. The release is available on Hugging Face and detailed in the company's launch announcement.

Kolibri-1 uses a mixture-of-experts (MoE) architecture, which activates only a subset of its parameters for any given token. That design has become a popular way to deliver stronger reasoning performance without paying the full compute cost of a dense model at inference time.

Why it matters

Aleph Alpha has positioned itself as a European champion in a field dominated by American and Chinese labs, and the "sovereign" framing speaks directly to that pitch. For public-sector and regulated industries in Europe, the appeal is concrete:

  • Open weights that can be run on-premises or within controlled infrastructure
  • First-class support for German alongside English
  • A reasoning-focused model rather than a general chat assistant

The emphasis on native German capability is notable. Many leading open models treat European languages as an afterthought, and a model tuned for German reasoning could find a receptive audience among organizations with data-residency and language requirements.

Aleph Alpha has not published detailed specifications such as parameter counts or context length in this record, so practical evaluation will depend on hands-on testing. Still, the arrival of another credible open-weight reasoning model from a European lab adds welcome diversity to a crowded but geographically narrow landscape.

Sources

  • Aleph-Alpha/Kolibri-1

    Hugging Face

    Visit
  • Kolibri: A Sovereign Open-Weight Model

    Hacker News

    Visit
OlderPerplexity releases a 27B model for multimodal routingPerplexity Ai · Vision-Language · 2 months agoNewerAllen Institute Open-Sources AstaBrief Report ModelAllen Institute for AI · Text / LLM · 2 months ago

Get the model

Hugging FaceHacker News

Specs

Context window262K tokens
Size78.8 GB
PrecisionFP8
ArchitectureKolibri1ForCausalLM
LicenseOTHER
Downloads9.6K
Likes1K

Can you run it?

Needs a workstation GPU (64 GB) at 4-bit.

  • FP8 (as published)

    H200 141 GB or 2×80 GB · 192 GB Mac Studio

    80.6 GB
  • 4-bit

    80 GB GPU (A100 / H100) · 128 GB Mac Studio

    48.7 GB
Your machine
GGUF builds
  • Mixture-of-experts: every expert must be loaded, so memory follows total parameters, not the active slice.
  • KV cache sized for 8,192 tokens of context; longer prompts need proportionally more.

Based on Hugging Face weights metadata. Assumes an 8K context and ~1 GB runtime overhead; actual needs vary. How we estimate


Modalities

ReasoningText / LLM

The Weekly Weights

Every open release that mattered, one email a week.

0 comments

No comments yet. Be the first to weigh in.

More from Aleph Alpha

All Aleph Alpha releases →
Aleph Alpha/Text / LLM

Aleph Alpha releases Kolibri, a sovereign open model

The German AI firm positions its new open-weight language model as a European alternative built with digital sovereignty in mind.

Oct 3, 2026

More in Reasoning

All Reasoning →
Sage S2.1 Flash
Internlm/ReasoningMulti-GPU server

InternLM's Sage S2.1 Flash targets scientific reasoning

The 320B mixture-of-experts model is built to reason over scientific evidence and act inside interactive environments.

Nov 15, 2026
ServiceNow/Text → SpeechConsumer GPU

ServiceNow Debuts Apriel Voice 27B, an Open Speech Model

The 27-billion-parameter preview unifies text-to-speech, transcription, and reasoning in a single open model.

Oct 28, 2026
SenseNova-Loop-9B
SenseTime/Vision-LanguageRuns on a laptop

SenseTime previews a 9B looped-transformer VLM

SenseNova-Loop-9B-Preview puts a recurrent-depth architecture to work on multimodal reasoning.

Oct 28, 2026