Qwen3.6 35B A3B qwen3.6-35b-a3b

qwen · qwen hybrid-reasoning efficient

#16 overall

params
36B (A3.3B)
arch
moe
context
256k
license
apache-2.0 open weights
released
Apr 2026
reasoning
yes
downloads/30d
6.2M

# why ranked

Overall: #16 — score 38.5 ● medium — one signal missing, weights renormalized

signalweightinput (0–100)
aa_intelligence 0.50 31.6
arena_elo 0.30
bench_composite 0.20 55.6

missing signals are dropped and the remaining weights renormalized — never imputed.

Coding: provisional — score 41.4 ● low — a single signal; treat with caution

provisional: too few independent signals for a numbered position — the score is shown, but this model sorts after every ranked model.

signalweightinput (0–100)
aa_coding 0.25 41.4
aider_polyglot 0.25
arena_elo_code 0.15
swe_bench_verified 0.35

missing signals are dropped and the remaining weights renormalized — never imputed.

full methodology

# trend

collecting history — trends appear after a few daily runs.

# benchmarks

benchmarkscoresourcedate
AA Coding Index 41.9 Artificial Analysis
AA Intelligence Index 31.6 Artificial Analysis
Ai2d 0.9 / 1 LLM Stats
AIME 2026 0.9 / 1 LLM Stats
C-eval 0.9 / 1 LLM Stats
Cc-ocr 0.8 / 1 LLM Stats
Charxiv-r 0.8 / 1 LLM Stats
Claw-eval 0.5 / 1 LLM Stats
Deep-planning 0.3 / 1 LLM Stats
Gdpval-aa 1056.0 / 3000 LLM Stats

# where to run

providerquantctx$/M in$/M out$/M blendtpsuptime
AkashMLfp8 256k $0.14$1.00$0.35 100.0%
AtlasCloudfp8 256k $0.16$0.97$0.36 100.0%
Parasailfp8 256k $0.15$1.00$0.36 100.0%
WandBfp8 256k $0.25$1.25$0.50 100.0%
SiliconFlowfp8 256k $0.20$1.60$0.55 93.3%

blended = (3·input + output) / 4 per 1M tokens · cheapest first

source aliases
aa
qwen3.6-35b-a3b
openrouter
qwen/qwen3.6-35b-a3b