The Known Good Updated 25 Jul 2026
Providers· First-party endpoint· 45 tracked models

Alibaba — pricing & performance

Compare models All providers Data sources
Models served
45
of 908 tracked models
Mean output speed
63
tokens per second, averaged over
41 measured endpoints
Mean latency
0.9s
time to first token, averaged over
41 measured endpoints
Cheapest blended
$0.11
per 1M tokens at a 3:1
input:output blend
Output speed on Alibaba

Median output tokens per second for each model this host serves, derived from OpenRouter endpoint telemetry. Higher is better. We do not run these measurements ourselves.

12 of 45 models ▾ ☰⚙⇩ +Add model from specific provider
183
142
138
129
112
110
104
98
81
77
72
65
The Known Good
45 models on Alibaba
Full leaderboard →
Model Creator Input $/M Output $/M Blended Speed TTFT Context
DeepSeek: DeepSeek V3.2 DeepSeek $0.37 $1.11 $0.56 45 0.8s 131k
DeepSeek: DeepSeek V4 Flash DeepSeek $0.13 $0.27 $0.17 77 0.9s 1.00M
DeepSeek: DeepSeek V4 Pro DeepSeek $1.42 $2.83 $1.77 52 1.4s 1.00M
Qwen: Qwen Plus 0728 Alibaba $0.26 $0.78 $0.39 — — 1.00M
Qwen: Qwen Plus 0728 (thinking) Alibaba $0.26 $0.78 $0.39 81 0.5s 1.00M
Qwen: Qwen-Plus Alibaba $0.26 $0.78 $0.39 51 0.5s 1.00M
Qwen: Qwen3 14B Alibaba $0.23 $0.91 $0.40 65 0.2s 131k
Qwen: Qwen3 235B A22B Alibaba $0.46 $1.82 $0.80 43 0.6s 131k
Qwen: Qwen3 235B A22B Instruct 2507 Alibaba $0.15 $0.60 $0.26 27 0.6s 131k
Qwen: Qwen3 235B A22B Thinking 2507 Alibaba $0.15 $1.50 $0.49 72 0.6s 131k
Qwen: Qwen3 30B A3B Alibaba $0.13 $0.52 $0.23 104 0.5s 131k
Qwen: Qwen3 30B A3B Instruct 2507 Alibaba $0.13 $0.52 $0.23 47 0.3s 131k
Qwen: Qwen3 30B A3B Thinking 2507 Alibaba $0.13 $1.56 $0.49 142 0.3s 82k
Qwen: Qwen3 32B Alibaba $0.10 $0.42 $0.18 — — 131k
Qwen: Qwen3 8B Alibaba $0.12 $0.46 $0.20 50 0.7s 131k
Qwen: Qwen3 Coder 30B A3B Instruct Alibaba $0.29 $1.46 $0.58 52 0.9s 262k
Qwen: Qwen3 Coder 480B A35B Alibaba $0.97 $4.88 $1.95 — — 262k
Qwen: Qwen3 Coder Flash Alibaba $0.20 $0.97 $0.39 39 1.2s 1.00M
Qwen: Qwen3 Coder Next Alibaba $0.30 $1.50 $0.60 62 2.1s 262k
Qwen: Qwen3 Coder Plus Alibaba $0.65 $3.25 $1.30 24 1.4s 1.00M
Qwen: Qwen3 Max Alibaba $0.78 $3.90 $1.56 52 1.0s 262k
Qwen: Qwen3 Max Thinking Alibaba $0.78 $3.90 $1.56 — — 262k
Qwen: Qwen3 Next 80B A3B Instruct Alibaba $0.098 $0.78 $0.27 98 0.5s 131k
Qwen: Qwen3 Next 80B A3B Thinking Alibaba $0.098 $0.78 $0.27 183 0.7s 131k
Qwen: Qwen3 VL 235B A22B Instruct Alibaba $0.26 $1.04 $0.46 19 0.5s 131k
Qwen: Qwen3 VL 235B A22B Thinking Alibaba $0.26 $2.60 $0.84 58 0.9s 131k
Qwen: Qwen3 VL 30B A3B Instruct Alibaba $0.13 $0.52 $0.23 19 0.4s 131k
Qwen: Qwen3 VL 30B A3B Thinking Alibaba $0.13 $1.56 $0.49 112 1.2s 131k
Qwen: Qwen3 VL 32B Instruct Alibaba $0.10 $0.42 $0.18 26 1.0s 131k
Qwen: Qwen3 VL 8B Instruct Alibaba $0.12 $0.46 $0.20 56 0.4s 131k
Qwen: Qwen3 VL 8B Thinking Alibaba $0.12 $1.36 $0.43 138 0.5s 131k
Qwen: Qwen3.5 397B A17B Alibaba $0.39 $2.34 $0.88 45 0.9s 262k
Qwen: Qwen3.5 Plus 2026-02-15 Alibaba $0.26 $1.56 $0.58 54 1.5s 1.00M
Qwen: Qwen3.5 Plus 2026-04-20 Alibaba $0.30 $1.80 $0.68 45 1.6s 1.00M
Qwen: Qwen3.5-122B-A10B Alibaba $0.26 $2.08 $0.71 129 1.6s 262k
Qwen: Qwen3.5-27B Alibaba $0.20 $1.56 $0.54 23 1.0s 262k
Qwen: Qwen3.5-35B-A3B Alibaba $0.16 $1.30 $0.45 65 1.1s 262k
Qwen: Qwen3.5-Flash Alibaba $0.065 $0.26 $0.11 110 0.7s 1.00M
Qwen: Qwen3.6 27B Alibaba $0.45 $2.70 $1.01 45 0.9s 262k
Qwen: Qwen3.6 Flash Alibaba $0.19 $1.12 $0.42 64 0.9s 1.00M
Qwen: Qwen3.6 Max Preview Alibaba $1.04 $6.24 $2.34 46 1.4s 262k
Qwen: Qwen3.6 Plus Alibaba $0.33 $1.95 $0.73 53 0.8s 1.00M
Qwen: Qwen3.7 Max Alibaba $1.48 $4.42 $2.21 44 1.2s 1.00M
Qwen: Qwen3.7 Plus Alibaba $0.32 $1.28 $0.56 13 0.9s 1.00M
Z.ai: GLM 5.2 Z.ai $0.83 $2.60 $1.27 65 1.2s 1.05M
The Known Good
Source. Prices, endpoint availability and throughput for Alibaba are ingested from OpenRouter; model metadata is enriched from models.dev. A blank cell means the figure was not published for that endpoint — it never means zero. Blended price is (3 × input + output) ÷ 4, the same 3:1 blend used everywhere on this site; see methodology. Quality scores are not provider-specific: a model scores the same wherever it is hosted, so quality lives on the model pages.