modelbenchmark.io

Qwen3.5 35B-A3B

Alibaba · qwen3-5-35b-a3b

Compare

Qwen vision-language model for visual reasoning, documents, and agent tasks

Specification

alibaba listing

Context
262K
Max output
66K
Released
2026-02-23
Knowledge cutoff
2025-04
Retires
2026-10-05
Open weights
yes
Input
text, image, video
Output
text

Price

US dollars per million tokens · alibaba list price

Input
$0.25
Output
$2.00
Cache read
$0.05

Quality

2 benchmarks · 1 source

Composite
79th
percentile of 344 scored models
Rank
73 / 344
±0.09 sd
Evidence
2 × 1
too little evidence to place with confidence
Effort range
—
one configuration only
reasoning84th1/4 bench
math65th1/7 bench

Every score

one row per source and configuration — nothing averaged away

BenchmarkScoreConfigurationSourceRun
GPQA diamond82.3 ±2.3default · 2 runsEpoch AI2026-08-28
OTIS Mock AIME 2024-202562.2 ±6.0default · 2 runsEpoch AI2026-08-30

The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.

Available from 25 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
alibabavendor$0.25$2.00——262K66K—
aihubmix$0.0564$0.4512——262K66K—
empiriolabs$0.057$0.459$0.057—258K64K—
merge-gateway$0.057$0.459$0.0204—131K33K—
orcarouter$0.057$0.459——262K66K—
siliconflow-cn$0.0598$0.4782——262K66K—
302ai$0.06$0.46——262K66K—
deepinfra$0.14$1.00$0.05—262K
82K
262K
—
nano-gpt$0.14$1.00$0.05—260K66K—
nano-gpt · thinking$0.14$1.00$0.05—260K66K—
openrouter$0.14$1.00$0.05—262K82K—
requesty$0.14$1.00$0.05—262K66K—
venice$0.15$1.00$0.05—256K16K—
kilo$0.1625$1.30——262K82K—
siliconflow$0.24$1.80$0.15—262K262K—
huggingface$0.25$2.00——262K66K—
jalapeno$0.25$2.00——262K66K—
llmgateway$0.25$2.00——262K66K—
llmgateway-providers$0.25$2.00——262K66K—
mixlayer$0.25$1.30——262K262K—
novita$0.25$2.00——262K66K—
novita-ai$0.25$2.00——262K66K—
wandb · wandb$0.25$1.25——262K262K2026-10-05
ofox$0.29$1.83$0.29—256K64K—
ofox · bailian$0.29$1.83$0.29—262K66K—

Price history

append-only observations · a listing writes a row only when its price moves

DateHostField$/M
2026-09-02nano-gptCache read$0.1125
2026-10-19nano-gptCache read$0.05
2026-09-02nano-gpt · thinkingCache read$0.1125
2026-10-19nano-gpt · thinkingCache read$0.05
2026-09-02openrouterCache read$0.25
2026-09-06openrouterCache read$0.1562
2026-09-13openrouterCache read$0.1562
2026-09-15openrouterCache read$0.1562
2026-09-20openrouterCache read$0.1562
2026-09-28openrouterCache read$0.1562
2026-10-03openrouterCache read$0.05
2026-10-09openrouterCache read$0.05
2026-10-10openrouterCache read$0.04
2026-10-14openrouterCache read$0.05
2026-09-02veniceCache read$0.1562
2026-10-09veniceCache read$0.05
2026-09-02nano-gptInput$0.225
2026-10-19nano-gptInput$0.14
2026-09-02nano-gpt · thinkingInput$0.225
2026-10-19nano-gpt · thinkingInput$0.14
2026-09-02openrouterInput$0.25
2026-09-05openrouterInput$0.08
2026-09-06openrouterInput$0.3125
2026-09-13openrouterInput$0.3125
2026-09-15openrouterInput$0.1625
2026-09-17openrouterInput$0.1625
2026-09-20openrouterInput$0.3125
2026-09-28openrouterInput$0.1625
2026-10-03openrouterInput$0.15
2026-10-09openrouterInput$0.1625
2026-10-10openrouterInput$0.08
2026-10-14openrouterInput$0.15
2026-10-21openrouterInput$0.14
2026-09-02siliconflow-cnInput$0.23
2026-10-22siliconflow-cnInput$0.0598
2026-09-02veniceInput$0.3125
2026-10-09veniceInput$0.15
2026-09-02nano-gptOutput$1.80
2026-10-19nano-gptOutput$1.00
2026-09-02nano-gpt · thinkingOutput$1.80
2026-10-19nano-gpt · thinkingOutput$1.00
2026-09-02openrouterOutput$1.25
2026-09-05openrouterOutput$0.75
2026-09-06openrouterOutput$1.25
2026-09-15openrouterOutput$1.30
2026-09-17openrouterOutput$1.30
2026-09-20openrouterOutput$1.25
2026-09-28openrouterOutput$1.30
2026-10-03openrouterOutput$1.00
2026-10-09openrouterOutput$1.30
2026-10-10openrouterOutput$0.75
2026-10-14openrouterOutput$1.00
2026-09-02siliconflow-cnOutput$1.86
2026-10-22siliconflow-cnOutput$0.4782
2026-09-02veniceOutput$1.25
2026-10-09veniceOutput$1.00

More from Alibaba

most-hosted first