FINAL-Bench/Darwin-4B-Opus
Text Generation ā¢ Updated ā¢ 12
Liu
Akicou
AI & ML interests
Check out Akicou/aboutMe
Recent Activity
liked a model about 3 hours ago
FINAL-Bench/Darwin-4B-Opus liked a model about 3 hours ago
arsovskidev/Gemma-4-E4B-Claude-4.6-Opus-Reasoning-Distilled liked a dataset about 3 hours ago
glyphsoftware/opus-4.6-frontend-developmentOrganizations
liked 2
models about 3 hours ago
liked a
dataset about 3 hours ago
updated a
model 3 days ago
liked a
dataset 3 days ago
Where's Steel Ball Run, Netflix?
š„ 10
2
#2 opened 5 days ago
by 
Alterious
love these
#5 opened 5 days ago
by Akicou
liked 2
models 6 days ago
š© Report: Copyright infringement
1
#212 opened 25 days ago
by 
radice
Where GLM-5.1?
ā 14
9
#72 opened 10 days ago
by 
DedeProGames
published a
model 7 days ago
reactedto SeaWolf-AI's post with š„ 7 days ago
Post
2140
š§¬ Darwin-35B-A3B-Opus ā The Child That Surpassed Both Parents
What if a merged model could beat both its parents? We proved it can.
Darwin-35B-A3B-Opus is a 35B MoE model (3B active) built with our Darwin V5 engine ā the first evolution system that CT-scans parent models before merging them.
š¤ Model: FINAL-Bench/Darwin-35B-A3B-Opus
The result speaks for itself: GPQA Diamond 90.0%, versus Father (Qwen3.5-35B-A3B) at 84.2% and Mother (Claude 4.6 Opus Distilled) at 85.0%. That's +6.9% over Father and +5.9% over Mother. Not a tradeoff ā a genuine leap. Meanwhile, MMMLU sits at 85.0% (Father: 85.2%), multimodal is fully intact, and all 201 languages are preserved.
How? Model MRI changed everything. Traditional merging is guesswork. Darwin V4 added evolution. Darwin V5 added X-ray vision. Model MRI scans each parent layer by layer and discovers: Mother's L34āL38 is the reasoning engine (peak cosine distance), 50ā65% of Mother's experts are dead (killed by text-only distillation), and Father is a healthy generalist with every expert alive. The prescription: transplant Mother's reasoning brain at L38 (90% weight), replace her dead experts with Father's living ones, and let Father's router handle the output layer. Reasoning went up. Versatility stayed intact. No tradeoff ā just evolution.
35B total, 3B active (MoE) Ā· GPQA Diamond 90.0% Ā· MMMLU 85.0% (201 languages) Ā· Multimodal Image & Video Ā· 262K native context Ā· 147.8 tok/s on H100 Ā· Runs on a single RTX 4090 (Q4) Ā· Apache 2.0
Darwin V5's full algorithm and technical details will be released alongside an upcoming paper.
š Live Demo: FINAL-Bench/Darwin-35B-A3B-Opus
š FINAL Bench Leaderboard: FINAL-Bench/Leaderboard
š ALL Bench Leaderboard: FINAL-Bench/all-bench-leaderboard
Built by VIDRAFT Ā· Supported by the Korean Government GPU Support Program
What if a merged model could beat both its parents? We proved it can.
Darwin-35B-A3B-Opus is a 35B MoE model (3B active) built with our Darwin V5 engine ā the first evolution system that CT-scans parent models before merging them.
š¤ Model: FINAL-Bench/Darwin-35B-A3B-Opus
The result speaks for itself: GPQA Diamond 90.0%, versus Father (Qwen3.5-35B-A3B) at 84.2% and Mother (Claude 4.6 Opus Distilled) at 85.0%. That's +6.9% over Father and +5.9% over Mother. Not a tradeoff ā a genuine leap. Meanwhile, MMMLU sits at 85.0% (Father: 85.2%), multimodal is fully intact, and all 201 languages are preserved.
How? Model MRI changed everything. Traditional merging is guesswork. Darwin V4 added evolution. Darwin V5 added X-ray vision. Model MRI scans each parent layer by layer and discovers: Mother's L34āL38 is the reasoning engine (peak cosine distance), 50ā65% of Mother's experts are dead (killed by text-only distillation), and Father is a healthy generalist with every expert alive. The prescription: transplant Mother's reasoning brain at L38 (90% weight), replace her dead experts with Father's living ones, and let Father's router handle the output layer. Reasoning went up. Versatility stayed intact. No tradeoff ā just evolution.
35B total, 3B active (MoE) Ā· GPQA Diamond 90.0% Ā· MMMLU 85.0% (201 languages) Ā· Multimodal Image & Video Ā· 262K native context Ā· 147.8 tok/s on H100 Ā· Runs on a single RTX 4090 (Q4) Ā· Apache 2.0
Darwin V5's full algorithm and technical details will be released alongside an upcoming paper.
š Live Demo: FINAL-Bench/Darwin-35B-A3B-Opus
š FINAL Bench Leaderboard: FINAL-Bench/Leaderboard
š ALL Bench Leaderboard: FINAL-Bench/all-bench-leaderboard
Built by VIDRAFT Ā· Supported by the Korean Government GPU Support Program
- 8 replies
liked a
model 8 days ago
liked a
model 10 days ago
liked a
model 13 days ago
liked a
dataset 14 days ago
liked a
model 14 days ago
liked a
model 16 days ago
liked a
model 26 days ago
