The questions on this site were asked by AI language models; the explanatory texts and translations were also written by AI models. Who checked what is shown alongside. → About the questions
41 models & run variants6 API variants6 Colab variants2 studies41 models & run variants
AMANi DUNiA · The models
Decision Models
Models, run status and their answers per report – updated automatically from answer files.
A growing model catalogue: what is already in the studies, which runs have been reported, what may be tested next – and what each model answered for each report.
Update dated · 2026-10-30
Run states and new technical specifications are a supplied working update, not a completed evaluation. Variants of the same model can appear in more than one environment.
Model catalogue updated: 2026-11-04T15:20:55+00:00
Answers per report
9,247 archive entries with all available model outputs. Entries without model answers are included. Available AI summaries appear in the detail view. Missing data is not a substantive answer.
Loading the list of reports …
Answers as of: Study 2, 2026-10-30; other models from the import of 2026-11-03.
Questions version v4.6
41 entries
No matching models.
From the studies · 17
Cloudflareevaluated
clef:27b
Type
Decision model
Size
27B
Study
2
AA intelligence – · code – · agent –
Partial · 8,678 of 9,247 entries with all 35 answers.
Reported complete for all 2,253 reports (follow-up run 1 Nov 2026 added the 384 DeepMind reports). Technical specifications come from the run registration.
Partial · 8,602 of 9,247 entries with all 35 answers.
Reported complete for all 2,253 reports (follow-up run 1 Nov 2026 added the missing reports and the 384 DeepMind reports). The text in the state is shortened to 32,000 characters.
Partial · 6,402 of 9,247 entries with all 35 answers.
Reported environment: T4 with Ollama. Token context sizes differ from the additional 64 KiB request limit for text plus questions. Necessary shortening is recorded in the run.
Cloudflarerunning
clef-flash:9b
Type
Decision model
Size
9B
Context / text
64K tokens · context
AA intelligence – · code – · agent –
Fallback to 32K or 16K is possible; the actual value is recorded.
Partial · 2,253 of 9,247 entries with all 35 answers.
As for the language models: AA scores from this site’s data (status of the language models). Only models AA lists under exactly this name; otherwise “–”. This site’s data does not yet hold code or agent scores for any model.
Model
Provider
AA intelligence
AA code
AA agent
as of
clef:27b
Cloudflare
–
–
–
–
Jev 1.14.0
TypeSafe
–
–
–
–
GPT-6 Luna
OpenAI
37
–
–
2026-10-01
Jev 1.13
TypeSafe
–
–
–
–
Clef Flash 9B
Cloudflare
–
–
–
–
Nimble 9B
Bespoke Labs
–
–
–
–
Tev1 4B
Together AI
–
–
–
–
Tev1 0.8B
Together AI
–
–
–
–
Laya 0.3.25
ConvAI
–
–
–
–
Drex Max v1.1
Nace.AI
–
–
–
–
d3-ultra
vLLM Semantic Router
–
–
–
–
d3-flash
vLLM Semantic Router
–
–
–
–
Mercury Decide
Inception Labs
–
–
–
–
amos-1-large
Scira · Amos
–
–
–
–
clef-flash:9b
Cloudflare
–
–
–
–
kev:9b-q4_K_M
–
–
–
–
–
nimble:9b
–
–
–
–
–
d1-omni:600m
–
–
–
–
–
nimble-v3:9b
–
–
–
–
–
d1:3b
–
–
–
–
–
d1-ultra
Liquid AI
–
–
–
–
d1.1
Liquid AI
–
–
–
–
clef-flash
Cloudflare
–
–
–
–
clef-omni
Cloudflare
–
–
–
–
joey-1
Baseten
–
–
–
–
elephant-1
Baseten
–
–
–
–
mammoth-1
Baseten
–
–
–
–
d3
–
–
–
–
–
tev1
–
–
–
–
–
laya
–
–
–
–
–
Ollaya
–
–
–
–
–
strands-decider:2b
–
–
–
–
–
subq-snap-1.1
Subq.ai
–
–
–
–
GLM-5.3-Flash
Z.ai
42
–
–
2026-10-01
Claude Haiku 5.5 · high
Anthropic
–
–
–
–
Claude Haiku 5.5 · max
Anthropic
–
–
–
–
d3:2b · lokal
vLLM Semantic Router
–
–
–
–
Claude Haiku 5.5 · low
Anthropic
–
–
–
–
kev:9b-q4_K_M · Kaggle
–
–
–
–
–
nimble-v3:9b · Kaggle
–
–
–
–
–
nimble:9b · Kaggle · Batch 5
–
–
–
–
–
Preprocessing, not a decision model
OpenDataLab
MinerU 4.1.3
Prepares PDFs as text before classification; not a decision model.
Data and interpretation
Completed results remain in the individual studies. New status reports are dated interim updates.