Model
Model explorer

Luminous Supreme (70B)

CLOSED
Aleph Alpha · Luminous family · released Apr 14, 2022

Germany's first large-scale sovereign LLM; API-only 70B flagship of the original Luminous family, now a legacy offering behind PhariaAI.

ReasoningCodingVisionFunction callingTool useAgentic
-190.2
Elo · rank #424
Parameters
70B
Active params
70B (dense)
Context
2K tokens
Architecture
Dense decoder-only autoregressive transformer with rotary position embeddings
License
Proprietary (Aleph Alpha API Terms & Conditions, SaaS-only)
Languages
5+
API price (in/out)
$175 / $192.5
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
ARC-ChallengeReasoning44.2%#139
best: Llama 3.1 405B · 96.9%
ARC-EasyReasoning76.9%#32
best: Phi-3-medium (14B) · 97.7%
HellaSwagReasoning59.2%#142
best: Claude 3 Opus · 95.4%
LAMBADAReasoning74.0%#12
best: GPT-3 175B · 86.4%
OpenBookQAReasoning33.0%#49
best: Claude 1 · 90.8%
PIQAReasoning79.9%#44
best: GPT-4o mini · 93.1%
RACE-HReasoning40.8%#20
best: Claude 3 Opus · 92.9%
TriviaQAKnowledge37.7%#45
best: Sarvam-1 (2B) · 90.6%
WinoGrandeReasoning70.0%#104
best: PaLM 2 · 90.9%
Run it locally
Closed weights — available via API only. No local deployment.
Input / M tok
$175
Output / M tok
$192.5
Luminous family
Elo progression across releases
API price $175/$192.5 · each benchmark row carries its own source badge (see methodology)