Alpha
Contact
Use · Soofi

Is it good enough?

Ownership levelPartialnone·limited·partial·substantial·fullAnalytical input C · 63.2/100

This page is a projection of the one entry record, the Reliability factor that Use covers. The full verdict is set by all four factors together, floor-weighted so the weakest caps the whole.

Which domain expands which factor
  • AssessUse & modify + Transparency
  • ImplementData control + Reliability
  • UseReliability
  • SupportTransparency

Capabilities & modalities

Soofi-S is a text-only language model for German and English generation, code, and technical and agentic tasks in industrial and regulated settings. Per the pretraining report and independent coverage (ev-arxiv, ev-perf), it is the strongest "fully open" model on aggregate English and German benchmarks (a benchmark claim; the weights are gated, see Assess), ahead of OLMo 3 32B, Apertus 70B, EuroLLM 22B and Alia 40B. It also wins the code aggregates in both languages among open base models, while activating only ~3B of 30B parameters per token. Choose it when European sovereignty, auditability, German-language strength and long-context economics matter. It is a leading open-class model, not a frontier-vs-closed contender.

Context window & long-context behaviour

Soofi-S supports contexts up to 256k tokens, and its hybrid Mamba-2 design keeps decode throughput flat from 4k to 256k (a near-constant inference cache). That helps long-document and high-concurrency workloads. Coverage is partial because effective long-context recall at full length is not independently benchmarked. Treat 256k as the architectural ceiling, not a verified quality guarantee at full length.

Prompt format & chat template

The Instruct-Preview variant ships an embedded Jinja chat template; its default system prompt carries only a knowledge cutoff and date, with no identity text. Apply it instead of hand-rolling role markers:

messages = [
    {"role": "system", "content": "Du bist ein hilfreicher Assistent."},
    {"role": "user", "content": "Fasse das folgende Dokument zusammen ..."},
]
prompt = tok.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)

With GGUF, run llama-server --jinja so the embedded template (identity and tools) is applied verbatim. The base checkpoint has no chat template; it is a plain text-completion and fine-tuning model and should not be prompted as a chat assistant.

Language coverage

Soofi-S is a bilingual German and English model. German was deliberately up-weighted in the ~27T-token training mix, and both languages are first-class supported targets, as a core design goal. Other languages may appear but are not supported targets. Evaluate per language before relying on Soofi-S outside German and English.

Function / tool calling

Soofi-S has a native tool-call format embedded in the Instruct GGUF's Jinja template. To use it reliably, run under llama-server --jinja, which applies the model's own identity-and-tools template verbatim. Ollama's template intentionally omits the tool portion, so tool calling there falls back to prompt scaffolding. Coverage is partial for this reason: the capability exists and is documented, but behaviour depends on the runtime template.

How this scores

The ownership factor this domain covers, drawn from the one entry record.

3

ReliabilityIs it reliable and good enough for the job?

Moderate

Pre-release: the weights are available only through manual approval, so runtime dependability is unproven for most adopters and the documented-incomplete safety cannot yet be exercised. The reported ~3B-active-parameter strength counts as a capability point in the AOI score, not as an ownership factor.

How this scores (AOI sub-dimensions)
Operational4/5how practical it is to run, serve and maintain in productionRuns on mainstream stacks (HF transformers with trust_remote_code, vLLM, llama.cpp/llama-server and Ollama) with first-party GGUF and 3-bit quantizations and a near-constant long-context cache.
Safety3/5whether misuse risks are evaluated and guardrails are providedA post-trained Instruct-Preview variant exists and the L3S Research Centre contributes safety and evaluation frameworks.
What this means for adoptionToday you do not own the use of these weights. They are manually gated in a closed beta, the card says it is not an open release, and no licence is in force, so no published text grants use, modification or commercial use. What exists is thorough documentation: a pretraining report with full per-source data accounting and a released training codebase. A stated OSAID 1.0 commitment has no date. The marketing says open-source; the binding text says closed beta. Treat openness as promised, not delivered, and re-assess when weights and a licence are published.

Sources

The same evidence records as the entry sheet. Read means the text was verified; unverified means it is known to exist but not yet read.

Technical_reportread2026-07-25
The arXiv report ("A Sovereign, Open-Source Foundation Model for German and English"; PDF "Soofi S Pretraining Report v1.0") describes a ~30B (30B-A3B) hybrid Mamba-2/Transformer MoE.
Licenceread2026-11-19
The Soofi-S-Base Hugging Face page and Hub API show the repo gated "manual" and the licence tag "closed-beta" (license: other).
Model cardread2026-11-19
The Soofi-Project Hugging Face org lists 20 model repos, all gated "manual": the base, a Megatron-format base checkpoint (Soofi-S-Base-Megatron, created 2026-08-23), the Instruct, Isar and Rhine previews, and first-party GGUF, FP8 and EntQuant 2/3/4-bit builds of each preview.
Documentationread2026-11-19
The public soofi-project/Soofi-Pretraining GitHub repo (based on the Nvidia Nemotron 3 Nano architecture; last push 2026-07-15) says it holds "the complete codebase used for training Soofi S" (preprocessing, pretraining, midtraining).
Vendor announcementread2026-11-19
The project site soofi.info says "Soofi S befindet sich derzeit in der Beta- und Entwicklungsphase" and "Ein öffentlicher Open-Source-Release ist geplant", with no release date, no licence and no access-request process described.
Vendor announcementunverified2026-07-25
A consortium press release presents Soofi as a German public-private consortium project for sovereign industrial AI, trained on Deutsche Telekom's Industrial AI Cloud in Munich, coordinated by the KI Bundesverband and funded by the BMWE.
Third-party analysisunverified2026-07-25
CAIRNE's launch page describes the SOOFI consortium, its members (including CAIRNE Gold member L3S Research Centre) and its sovereign open-source foundation-model mission.
Third-party analysisunverified2026-07-25
Independent technology coverage reports Soofi-S as a leading fully-open model that meets the Open Source AI Definition, publishing weights, intermediate checkpoints, training and eval code and a detailed data inventory.
Third-party analysisunverified2026-07-25
Independent coverage reports Soofi-S is the strongest fully-open model on aggregate English and German benchmarks, ahead of OLMo 3 32B, Apertus 70B, EuroLLM 22B and Alia 40B, and winning code aggregates among open base models.