Qwen3.5 397B A17B qwen3.5-397b-a17b

qwen · qwen hybrid-reasoning flagship

#7 overall

params
403B (A17B)
arch
moe
context
250k
license
apache-2.0 open weights
released
Feb 2026
reasoning
yes
downloads/30d
416.1k

# why ranked

Overall: #7 — score 45.5 ● medium — one signal missing, weights renormalized

signalweightinput (0–100)
aa_intelligence 0.50 33.7
arena_elo 0.30
bench_composite 0.20 75.0

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

Coding: provisional — score 65.5 ● 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 65.5
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 48.2 Artificial Analysis
AA Intelligence Index 33.7 Artificial Analysis
Aa-lcr 0.7 / 1 LLM Stats
AIME 2026 0.9 / 1 LLM Stats
Bfcl-v4 0.7 / 1 LLM Stats
Browsecomp 0.7 / 1 LLM Stats
Browsecomp-zh 0.7 / 1 LLM Stats
C-eval 0.9 / 1 LLM Stats
Deep-planning 0.3 / 1 LLM Stats
Gdpval-aa 961.0 / 3000 LLM Stats

# where to run

providerquantctx$/M in$/M out$/M blendtpsuptime
Alibabaunknown 256k $0.39$2.34$0.88 100.0%
DigitalOceanunknown 128k $0.39$2.45$0.90 100.0%
Chutesfp8 256k $0.45$3.00$1.09 98.9%
DeepInfrafp8 256k $0.45$3.00$1.09 97.8%
Morphunknown 256k $0.50$3.50$1.25 100.0%
Parasailfp8 256k $0.50$3.60$1.27 100.0%
AtlasCloudfp8 256k $0.55$3.50$1.29 100.0%
Phalaunknown 256k $0.55$3.50$1.29 99.6%
GMICloudfp8 256k $0.60$3.60$1.35
Novitaunknown 256k $0.60$3.60$1.35
StreamLakeunknown 250k $0.60$3.60$1.35
Veniceunknown 125k $0.75$4.50$1.69

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

source aliases
aa
qwen3.5-397b-a17b
epoch
Qwen3.5 397B-A17B
openrouter
qwen/qwen3.5-397b-a17b