builds models · serves endpoints · 2026-08-14

Alibaba / Qwen

Alibaba / Qwen sells 2 different plans, easy to confuse. The Coding Plan is for agentic coding only and is metered in requests; the Token Plan is metered in credits and covers every modality. Pick the Coding Plan if you live in a coding agent and want a flat monthly cap; pick the Token Plan (Individual for one person, Team for seats) if you also do image/video or want the cheaper $6-$68 entry points. Neither converts cleanly to tokens. And mind which link you sign up through: the promotional price (50% new-user coupon, cheaper Team seats) only applies from specific campaign pages, so the same plan can cost different amounts by the door you enter. One non-price difference matters most: Alibaba lists "no model training of your content" as a Team-only feature, so on the cheaper Individual tiers your interactions can be used to train models.

Jump to: Coding Plan (Pro) (requests · coding only) · Token Plan (Individual + Team) (credits · all modalities)

Coding Plan (Pro)

Flat monthly cap for agentic coding, metered in requests — not tokens. One live tier: Pro at $50/month.

Buy — Model Studio →

Tier	$/month	Included usage
Pro	$50.00	—

console only partly convertible

As advertised. 6,000 requests / 5h · 45,000 / week · 90,000 / month

What that actually means. Request caps are published (6,000/5h) but no per-request cost, so requests do not convert to tokens. Feature request to show remaining was closed as not planned. A 50% new-user discount applies from the promo landing page but not from the plain docs link.

Where to check what is left. home.qwencloud.com/billing/coding-plan

Models covered. qwen3-coder-plus, qwen3-coder-next, qwen3.7-plus, qwen3.6-plus, qwen3.5-plus, qwen3-max, kimi-k2.5, glm-5, glm-4.7, MiniMax-M2.5

A SEPARATE product from the Token Plan below, and easy to confuse with it. The Coding Plan is agentic-coding only (Claude Code / Qwen Code / Cline style), metered in REQUESTS, and now has a single live tier: Pro at $50/month. The Lite tier was discontinued for new subs 2026-03-20 (renewals ended 2026-04-13), though comparison sites still list it. Docs say a coding task uses ~5-10 requests (simple) to 30+ (complex), so requests are not tasks and do not convert to tokens.

checked 2026-08-13 · high · source

Token Plan (Individual + Team)

Credits across every modality (text/image/video/audio). Buy it as an Individual (Lite/Standard/Pro, $6–$68/mo) or per Team seat. Credits do not convert to tokens — the exchange rate is unpublished.

Token Plan — Individual (Lite $6, 50% new-user coupon) → · Token Plan — Team (seats $20/$75/$200, cheaper than the $30/$100/$200 door) →

Tier	$/month	Included quota	Concurrent
Individual Lite	$6.00	~10,000	1-2
Individual Standard	$20.00	~40,000	3-4
Individual Pro	$70.00	~160,000	6-8
Team Standard seat	$20-30	25,000/seat	—
Team Pro seat	$75-100	100,000/seat	—
Team Max seat	$200.00	250,000/seat	—

console only units unstated

As advertised. Individual credits: Lite ~10,000/mo (2,500 per 7d, 700 per 5h) · Standard ~40,000/mo (10,000/7d, 3,000/5h) · Pro ~160,000/mo (40,000/7d, 12,000/5h). Team per seat: Standard 25,000/mo · Pro 100,000/mo · Max 250,000/mo. Concurrent agents: 1-2 / 3-4 / 6-8 (Individual).

What that actually means. Metered in CREDITS with a deliberately unpublished exchange rate. Neither Alibaba page states how many tokens a credit buys, a per-model credit-consumption rate, or a dollar value for a credit -- an independent decode concluded the plan is "priced in a currency whose exchange rate is unpublished", so it cannot be turned into a token quota. The credits are metered on rolling 5h and 7-day windows as well as monthly (so the monthly figure is a ceiling you cannot spend in a burst). A limited-time promo gives Qwen3.8-Max 10x usage per credit, which only underlines that a hidden per-credit usage rate exists. Distinct from the request-metered Coding Plan above: different product, different unit, all modalities (text/image/video/audio) not just coding. And the price depends on WHICH DOOR you enter: the two campaign pages quote different figures for the same Team seats ($20/$75/$200 vs $30/$100/$200), a 50% new-user coupon and a $2-off coupon apply only from specific links, and a limited-time promo can 10x usage per credit — so the same plan costs materially different amounts depending on the URL you sign up through.

Where to check what is left. alibabacloud.com model studio console

Models covered. qwen3.8-max-preview, qwen3.7-max, qwen3.7-plus, qwen3.6-flash, glm-5.2, deepseek-v4-pro, kimi, MiniMax-M2.5, (+ Wan / HappyHorse / Qwen-image: image/video/audio)

Individual vs Team — read before the price. Individual and Team are not just a price difference. Alibaba's own Individual-vs-Team comparison lists FOUR things as Team-only: enterprise-grade data security, enterprise management, unified agent/usage support, and — the one that matters most — "no model training of your content". Listed as a Team benefit, it means the Individual (Personal) plans carry no such guarantee: on the cheap $6-$68 tiers, your interactions can be used to train models. Team also adds a shared quota pack across seats. So the real trade is privacy and pooled quota (Team, from $20/seat) versus lowest entry price (Individual, from $6) — not just capacity.

checked 2026-08-13 · high (tiers/credits as listed) / the credit unit is undefined by the vendor · source 1 source 2 source 3

Models it builds

Model	Coding index	tok/s	$/M in	$/M out	$ / session	Cache
Qwen3.7 Max	66.0	0.0	$1.48	$4.42	$1.5594	95.1%
Qwen3.7 Plus	55.9	56.3	$0.32	$1.28	$0.5280	94.6%
Qwen3.6 Plus	54.5	0.0	$0.33	$1.95	—	—
Qwen3.6 27B	53.7	53.3	$0.60	$3.60	—	—
Qwen3.5 397B A17B	48.2	80.3	$0.39	$2.34	—	—
Qwen3.5-122B-A10B	45.7	132.8	$0.29	$2.40	—	—
Qwen3.6 35B A3B	41.9	138.3	$0.15	$1.00	—	—
Qwen3.5-35B-A3B	37.0	157.1	$0.22	$1.80	—	—
Qwen3 Coder Next	36.2	134.3	$0.12	$0.80	—	—
Qwen3.5-9B	28.7	84.6	$0.10	$0.15	—	—
Qwen3.8 Max	71.8	43.3	$2.00	$6.00	—	—
Qwen3 32B	15.3	0.0	$0.08	$0.28	—	—
Qwen3 14B	13.8	0.0	$0.12	$0.24	—	—
Qwen3 8B	9.0	0.0	$0.12	$0.45	—	—
Qwen3 235B A22B 2507	22.1	0.0	$0.45	$1.82	—	—
Qwen3 30B A3B 2507	12.1	0.0	$0.12	$0.50	—	—

Models it serves

18 endpoints, cheapest first. This is the view the model-by-model tables cannot give you: what one provider actually offers, and at what precision.

Model	Quant	$/M in	$/M out	Context	Uptime 30m
Qwen3 8B	unknown	$0.12	$0.45	131,072	100.0%
Qwen3 30B A3B 2507	unknown	$0.13	$0.52	131,072	100.0%
Qwen3 14B	unknown	$0.23	$0.91	131,072	100.0%
Qwen3.5-35B-A3B	unknown	$0.16	$1.30	262,144	99.4%
DeepSeek V3.2	fp8	$0.37	$1.11	131,072	96.9%
Qwen3.7 Plus	unknown	$0.32	$1.28	1,000,000	100.0%
Qwen3 Coder Next	unknown	$0.30	$1.50	262,144	—
Qwen3.5-122B-A10B	unknown	$0.26	$2.08	262,144	100.0%
Qwen3.6 Plus	unknown	$0.33	$1.95	1,000,000	100.0%
Qwen3 235B A22B 2507	unknown	$0.45	$1.82	131,072	100.0%
Qwen3.5 397B A17B	unknown	$0.39	$2.34	262,144	100.0%
Qwen3.6 27B	unknown	$0.45	$2.70	262,144	100.0%
GLM-5.2	fp8	$0.97	$3.04	1,048,576	100.0%
Kimi K2.7 Code	fp8	$0.95	$4.00	262,144	100.0%
GLM-5.1	fp8	$1.33	$4.18	202,745	99.5%
Qwen3.7 Max	unknown	$1.48	$4.42	1,000,000	100.0%
Qwen3.8 Max	unknown	$2.00	$6.00	1,000,000	100.0%
GLM-5.2	fp8	$2.31	$7.26	1,048,576	100.0%

Where to buy its models

Everyone serving these models, cheapest first, with the numeric precision each one runs. The lab is often not the cheapest place to buy its own model.

Model	Served by	Quant	$/M in	$/M out	Uptime 30m
Qwen3.7 Max	Alibaba / Qwen	unknown	$1.48	$4.42	100.0%
Qwen3.7 Plus	Alibaba / Qwen	unknown	$0.32	$1.28	100.0%
Qwen3.6 Plus	Alibaba / Qwen	unknown	$0.33	$1.95	100.0%
Qwen3.6 27B	Chutes	fp8	$0.30	$2.00	52.7%
Qwen3.6 27B	Morph	fp4	$0.29	$2.40	100.0%
Qwen3.6 27B	Phala	unknown	$0.32	$2.70	95.5%
Qwen3.6 27B	Alibaba / Qwen	unknown	$0.45	$2.70	100.0%
Qwen3.6 27B	Io Net	fp8	$0.39	$2.89	100.0%
Qwen3.6 27B	SiliconFlow	fp8	$0.30	$3.20	99.9%
Qwen3.6 27B	DeepInfra	fp8	$0.32	$3.20	99.9%
Qwen3.6 27B	Venice	fp8	$0.33	$3.25	100.0%
Qwen3.6 27B	CoreWeave	fp8	$0.60	$3.60	99.8%
Qwen3.5 397B A17B	DigitalOcean	unknown	$0.30	$1.93	100.0%
Qwen3.5 397B A17B	Alibaba / Qwen	unknown	$0.39	$2.34	100.0%
Qwen3.5 397B A17B	Chutes	fp8	$0.45	$3.00	99.7%
Qwen3.5 397B A17B	DeepInfra	fp8	$0.45	$3.00	99.8%
Qwen3.5 397B A17B	Parasail	fp8	$0.50	$3.60	99.9%
Qwen3.5 397B A17B	AtlasCloud	fp8	$0.55	$3.50	99.3%
Qwen3.5 397B A17B	Phala	unknown	$0.55	$3.50	100.0%
Qwen3.5 397B A17B	GMICloud	fp8	$0.60	$3.60	—
Qwen3.5 397B A17B	Novita	unknown	$0.60	$3.60	99.3%
Qwen3.5 397B A17B	StreamLake	unknown	$0.60	$3.60	96.2%
Qwen3.5 397B A17B	Venice	unknown	$0.75	$4.50	—
Qwen3.5-122B-A10B	Alibaba / Qwen	unknown	$0.26	$2.08	100.0%
Qwen3.5-122B-A10B	SiliconFlow	fp8	$0.26	$2.08	100.0%
Qwen3.5-122B-A10B	DeepInfra	fp4	$0.29	$2.40	99.7%
Qwen3.5-122B-A10B	AtlasCloud	fp8	$0.30	$2.40	99.9%
Qwen3.5-122B-A10B	Novita	bf16	$0.40	$3.20	99.8%
Qwen3.6 35B A3B	Venice	fp8	$0.10	$0.95	100.0%
Qwen3.6 35B A3B	DeepInfra	fp8	$0.10	$0.95	95.8%
Qwen3.6 35B A3B	AkashML	fp8	$0.14	$1.00	100.0%
Qwen3.6 35B A3B	Parasail	fp8	$0.15	$1.00	99.9%
Qwen3.6 35B A3B	AtlasCloud	fp8	$0.19	$1.11	100.0%
Qwen3.6 35B A3B	Io Net	fp8	$0.19	$1.19	100.0%
Qwen3.6 35B A3B	Phala	unknown	$0.20	$1.27	99.9%
Qwen3.6 35B A3B	CoreWeave	fp8	$0.25	$1.25	99.9%
Qwen3.6 35B A3B	SiliconFlow	fp8	$0.20	$1.60	99.9%
Qwen3.5-35B-A3B	AkashML	fp8	$0.14	$1.00	100.0%
Qwen3.5-35B-A3B	DeepInfra	fp8	$0.14	$1.00	100.0%
Qwen3.5-35B-A3B	Parasail	fp8	$0.15	$1.00	100.0%
Qwen3.5-35B-A3B	Alibaba / Qwen	unknown	$0.16	$1.30	99.4%
Qwen3.5-35B-A3B	CoreWeave	fp8	$0.25	$1.25	88.4%
Qwen3.5-35B-A3B	Venice	unknown	$0.31	$1.25	97.7%
Qwen3.5-35B-A3B	AtlasCloud	fp8	$0.22	$1.80	100.0%
Qwen3.5-35B-A3B	SiliconFlow	fp8	$0.24	$1.80	—
Qwen3 Coder Next	Parasail	bf16	$0.12	$0.80	100.0%
Qwen3 Coder Next	StreamLake	unknown	$0.18	$0.90	—
Qwen3 Coder Next	Novita	fp8	$0.20	$1.50	100.0%
Qwen3 Coder Next	Alibaba / Qwen	unknown	$0.30	$1.50	—
Qwen3.5-9B	DeepInfra	bf16	$0.10	$0.15	99.9%
Qwen3.5-9B	SiliconFlow	fp8	$0.10	$0.15	69.4%
Qwen3.5-9B	Venice	fp8	$0.10	$0.15	100.0%
Qwen3.5-9B	Parasail	bf16	$0.10	$0.25	100.0%
Qwen3.5-9B	Together	unknown	$0.17	$0.25	99.6%
Qwen3.8 Max	Alibaba / Qwen	unknown	$2.00	$6.00	100.0%
Qwen3 32B	DeepInfra	fp8	$0.08	$0.28	100.0%
Qwen3 32B	Nebius	fp8	$0.10	$0.30	99.8%
Qwen3 32B	SiliconFlow	fp8	$0.14	$0.57	100.0%
Qwen3 32B	Groq	unknown	$0.29	$0.59	100.0%
Qwen3 14B	NextBit	int4	$0.10	$0.22	99.3%
Qwen3 14B	DeepInfra	fp8	$0.12	$0.24	99.9%
Qwen3 14B	Alibaba / Qwen	unknown	$0.23	$0.91	100.0%
Qwen3 8B	Alibaba / Qwen	unknown	$0.12	$0.45	100.0%
Qwen3 235B A22B 2507	Alibaba / Qwen	unknown	$0.45	$1.82	100.0%
Qwen3 30B A3B 2507	DeepInfra	fp8	$0.12	$0.50	100.0%
Qwen3 30B A3B 2507	Alibaba / Qwen	unknown	$0.13	$0.52	100.0%

◆ Some links on this page are referral links: if you sign up through them this site may earn a commission, at no extra cost to you. Rankings, prices and measurements are taken from the sources listed under Method, and are not influenced by whether a provider has a referral programme.