BEAM

Models

Features

Rankings

Docs

New

Pro

Use Alibaba Models in Big-AGI.

Qwen3.8 Max and the rest of Model Studio at Alibaba's own rates - filtered to just the chat models, with resold models labeled "(Alibaba)".

Qwen3.8 27B DeepSeek V4 Pro 0813 (Alibaba)DeepSeek V4 Flash (Alibaba)DeepSeek V4 Pro (Alibaba)Kimi K3 (Alibaba)

Alibaba setup guide Alibaba models

All supported Alibaba models

ModelContextInputOutputReleased

Qwen3.8 27B

Open-weights 27B dense vision-language model of the Qwen3.8 line. 1M context, thinking, image/video input.

1M

$0.50

$3.00

Aug 2026

DeepSeek V4 Pro 0813 (Alibaba)

DeepSeek V4 Pro GA (0813) served via Alibaba Model Studio. Much stronger agentic and tool use than the April preview. 1M context, thinking.

1M

$2.40

$4.80

Aug 2026

Qwen3.8 2.4T-A95B

Open-weights release of the Qwen3.8 flagship: 2.4T sparse MoE, ~95B active. Text-only serving with 1M context and always-on thinking.

1M

$2.00

$6.00

Aug 2026

DeepSeek V4 Flash 0731 (Alibaba)

DeepSeek V4 Flash 0731 revision served via Alibaba Model Studio. 1M context, thinking.

1M

$0.20

$0.40

Jul 2026

Qwen3.8 Max

Flagship 2.4T-parameter sparse MoE multimodal model with 1M context, thinking, and vision/video understanding.

1M

$2.00

$6.00

Jul 2026

Kimi K3 (Alibaba)

Moonshot Kimi K3 flagship served via Alibaba Model Studio. Multimodal, thinking on by default, 1M context.

1M

$3.00

$15.00

Jul 2026

Qwen3.7 Flash

Latest fast multimodal model with 1M context, thinking (on by default), vision, and 128K output.

1M

$0.03

$0.13

Jul 2026

GLM-5.2 (Alibaba)

Zhipu GLM-5.2 served via Alibaba Model Studio. 1M context, thinking.

1M

$1.40

$4.40

Jun 2026

Kimi K2.7 Code (Alibaba)

Moonshot Kimi K2.7 Code served via Alibaba Model Studio. Multimodal, always-on thinking, 256K context.

262K

$0.95

$4.00

Jun 2026

Qwen3.7 Plus

Multimodal agent model with 1M context, native thinking, and vision/video understanding. Lower cost than Max.

1M

$0.40

$1.60

Jun 2026

Qwen3.7 Max

Flagship agent model with native extended thinking and 1M context. Text-only; strong at coding, productivity, and long-horizon autonomous tasks.

1M

$2.50

$7.50

May 2026

Qwen3.6 35B-A3B

Open 35B-A3B MoE multimodal with thinking. 256K context.

262K

$0.38

$2.25

Apr 2026

Qwen3.6 27B

Open 27B dense multimodal with thinking. 256K context.

262K

$0.60

$3.60

Apr 2026

Qwen3.6 Flash

Fast, cost-effective multimodal model with 1M context, near-flagship quality, vision/video, and built-in tools.

1M

$0.25

$1.50

Apr 2026

Qwen3.6 Max Preview

Qwen3.6 Max preview (never GA; superseded by Qwen3.7 Max). 256K context, thinking, text-only.

262K

$1.30

$7.80

Apr 2026

DeepSeek V4 Flash (Alibaba)

DeepSeek V4 Flash served via Alibaba Model Studio. 1M context, thinking.

1M

$0.20

$0.40

Apr 2026

DeepSeek V4 Pro (Alibaba)

DeepSeek V4 Pro served via Alibaba Model Studio (Alibaba pricing, well above DeepSeek-direct). 1M context, thinking.

1M

$2.40

$4.80

Apr 2026

GLM-5.1 (Alibaba)

Zhipu GLM-5.1 served via Alibaba Model Studio, superseded by GLM-5.2. 200K context, thinking.

203K

$1.40

$4.40

Apr 2026

Qwen3.6 Plus

Previous Plus tier, superseded by Qwen3.7 Plus. 1M context, thinking, vision.

1M

$0.50

$3.00

Apr 2026

Qwen3.5 122B-A10B

Open 122B-A10B MoE multimodal with thinking. 256K context.

262K

$0.40

$3.20

Feb 2026

Qwen3.5 35B-A3B

Open 35B-A3B MoE multimodal with thinking. 256K context.

262K

$0.25

$2.00

Feb 2026

Qwen3.5 27B

Open 27B dense multimodal with thinking. 256K context.

262K

$0.30

$2.40

Feb 2026

Qwen3.5 Flash

Former Flash tier, superseded by Qwen3.6/3.7 Flash. 1M context, thinking, vision.

1M

$0.10

$0.40

Feb 2026

Qwen3.5 397B-A17B

Open 397B-A17B MoE multimodal with thinking. 256K context.

262K

$0.60

$3.60

Feb 2026

Qwen3.5 Plus

Former Plus tier, superseded by Qwen3.6/3.7 Plus. 1M context, thinking, vision.

1M

$0.40

$2.40

Feb 2026

Qwen3 Coder Next

Budget agentic coder on the Qwen3-Next architecture. 256K context, non-thinking, tiered pricing.

262K

$0.30

$1.50

Feb 2026

DeepSeek V3.2 (Alibaba)

DeepSeek V3.2 served via Alibaba Model Studio (superseded by V4). Thinking.

131K

$0.57

$1.71

Dec 2025

Qwen3 VL Flash

Budget VL tier below Qwen3 VL Plus. 256K context, thinking, vision.

262K

$0.05

$0.40

Oct 2025

Qwen3 Vl 235b A22b Thinking

Alibaba model (not yet curated).

131K

-

-

Sep 2025

Qwen3 Vl 235b A22b Instruct

Alibaba model (not yet curated).

131K

-

-

Sep 2025

Qwen3 Coder Plus

Agentic coding model with very long context. Tiered pricing by input length (up to 1M).

1M

$1.00

$5.00

Sep 2025

Qwen3 Max

Retired 2025 flagship, superseded by Qwen3.6+ Max. 256K context; the base id now serves the thinking-capable 2026-01-23 snapshot.

262K

$1.20

$6.00

Sep 2025

Qwen3 Next 80b A3b Thinking

Alibaba model (not yet curated).

131K

-

-

Sep 2025

Qwen3 Next 80b A3b Instruct

Alibaba model (not yet curated).

131K

-

-

Sep 2025

Qwen3 30b A3b Thinking 2507

Alibaba model (not yet curated).

131K

-

-

Aug 2025

Qwen3 30b A3b Instruct 2507

Alibaba model (not yet curated).

131K

-

-

Jul 2025

Qwen3 Coder Flash

Former budget coder, superseded by Qwen3 Coder Next (which lacks its 1M context). Non-thinking.

1M

$0.30

$1.50

Jul 2025

Qwen3 235b A22b Thinking 2507

Alibaba model (not yet curated).

131K

-

-

Jul 2025

Qwen3 30b A3b

Alibaba model (not yet curated).

131K

-

-

Apr 2025

Qwen3 32b

Alibaba model (not yet curated).

131K

-

-

Apr 2025

Qwen3 14b

Alibaba model (not yet curated).

131K

-

-

Apr 2025

Qwen3 235b A22b

Alibaba model (not yet curated).

131K

-

-

Apr 2025

Qwen3 8b

Alibaba model (not yet curated).

131K

-

-

Apr 2025

Qwen Plus

Balanced quality, speed, and cost with hybrid thinking. 1M context.

1M

$0.40

$1.20

Feb 2025

Qwen3 235b A22b Instruct 2507

Alibaba model (not yet curated).

131K

-

-

-

GLM-5.2 Fast (Alibaba)

Zhipu GLM-5.2 fast-serving tier via Alibaba Model Studio (preview). Same model, lower latency, ~2x price.

1M

$2.80

$8.80

-

Qvq Max

Alibaba model (not yet curated).

131K

-

-

-

Qwen Coder Plus

Alibaba model (not yet curated).

131K

-

-

-

Qwen Flash

Fast and very low cost with hybrid thinking. 1M context.

1M

$0.05

$0.40

-

Qwen Max

Best quality of the stable commercial line. 32K context.

33K

$1.60

$6.40

-

Qwen Plus

Balanced quality, speed, and cost with hybrid thinking. 1M context.

1M

$0.40

$1.20

-

Qwen Turbo

Fastest and cheapest for simple tasks. 1M context.

1M

$0.05

$0.20

-

Qwen Vl Max

Alibaba model (not yet curated).

131K

-

-

-

Qwen Vl Plus

Alibaba model (not yet curated).

131K

-

-

-

Qwen3.7 Max

Flagship agent model with native extended thinking and 1M context. Text-only; strong at coding, productivity, and long-horizon autonomous tasks.

1M

$2.50

$7.50

-

Qwen3 Coder 480b A35b Instruct

Alibaba model (not yet curated).

131K

-

-

-

Qwen3 Max

Retired 2025 flagship, superseded by Qwen3.6+ Max. 256K context; the base id now serves the thinking-capable 2026-01-23 snapshot.

262K

$1.20

$6.00

-

Qwen3 VL Plus

Current vision-language model with strong visual reasoning and thinking. Tiered pricing by input length (up to 256K).

262K

$0.20

$1.60

-

Qwq Plus

Alibaba model (not yet curated).

131K

-

-

-

Qwen3.8 27B

Open-weights 27B dense vision-language model of the Qwen3.8 line. 1M context, thinking, image/video input.

1M · in $0.50 · out $3.00

DeepSeek V4 Pro 0813 (Alibaba)

DeepSeek V4 Pro GA (0813) served via Alibaba Model Studio. Much stronger agentic and tool use than the April preview. 1M context, thinking.

1M · in $2.40 · out $4.80

Qwen3.8 2.4T-A95B

Open-weights release of the Qwen3.8 flagship: 2.4T sparse MoE, ~95B active. Text-only serving with 1M context and always-on thinking.

1M · in $2.00 · out $6.00

DeepSeek V4 Flash 0731 (Alibaba)

DeepSeek V4 Flash 0731 revision served via Alibaba Model Studio. 1M context, thinking.

1M · in $0.20 · out $0.40

Qwen3.8 Max

Flagship 2.4T-parameter sparse MoE multimodal model with 1M context, thinking, and vision/video understanding.

1M · in $2.00 · out $6.00

Kimi K3 (Alibaba)

Moonshot Kimi K3 flagship served via Alibaba Model Studio. Multimodal, thinking on by default, 1M context.

1M · in $3.00 · out $15.00

Qwen3.7 Flash

Latest fast multimodal model with 1M context, thinking (on by default), vision, and 128K output.

1M · in $0.03 · out $0.13

GLM-5.2 (Alibaba)

Zhipu GLM-5.2 served via Alibaba Model Studio. 1M context, thinking.

1M · in $1.40 · out $4.40

Kimi K2.7 Code (Alibaba)

Moonshot Kimi K2.7 Code served via Alibaba Model Studio. Multimodal, always-on thinking, 256K context.

262K · in $0.95 · out $4.00

Qwen3.7 Plus

Multimodal agent model with 1M context, native thinking, and vision/video understanding. Lower cost than Max.

1M · in $0.40 · out $1.60

Qwen3.7 Max

Flagship agent model with native extended thinking and 1M context. Text-only; strong at coding, productivity, and long-horizon autonomous tasks.

1M · in $2.50 · out $7.50

Qwen3.6 35B-A3B

Open 35B-A3B MoE multimodal with thinking. 256K context.

262K · in $0.38 · out $2.25

Qwen3.6 27B

Open 27B dense multimodal with thinking. 256K context.

262K · in $0.60 · out $3.60

Qwen3.6 Flash

Fast, cost-effective multimodal model with 1M context, near-flagship quality, vision/video, and built-in tools.

1M · in $0.25 · out $1.50

Qwen3.6 Max Preview

Qwen3.6 Max preview (never GA; superseded by Qwen3.7 Max). 256K context, thinking, text-only.

262K · in $1.30 · out $7.80

DeepSeek V4 Flash (Alibaba)

DeepSeek V4 Flash served via Alibaba Model Studio. 1M context, thinking.

1M · in $0.20 · out $0.40

DeepSeek V4 Pro (Alibaba)

DeepSeek V4 Pro served via Alibaba Model Studio (Alibaba pricing, well above DeepSeek-direct). 1M context, thinking.

1M · in $2.40 · out $4.80

GLM-5.1 (Alibaba)

Zhipu GLM-5.1 served via Alibaba Model Studio, superseded by GLM-5.2. 200K context, thinking.

203K · in $1.40 · out $4.40

Qwen3.6 Plus

Previous Plus tier, superseded by Qwen3.7 Plus. 1M context, thinking, vision.

1M · in $0.50 · out $3.00

Qwen3.5 122B-A10B

Open 122B-A10B MoE multimodal with thinking. 256K context.

262K · in $0.40 · out $3.20

Qwen3.5 35B-A3B

Open 35B-A3B MoE multimodal with thinking. 256K context.

262K · in $0.25 · out $2.00

Qwen3.5 27B

Open 27B dense multimodal with thinking. 256K context.

262K · in $0.30 · out $2.40

Qwen3.5 Flash

Former Flash tier, superseded by Qwen3.6/3.7 Flash. 1M context, thinking, vision.

1M · in $0.10 · out $0.40

Qwen3.5 397B-A17B

Open 397B-A17B MoE multimodal with thinking. 256K context.

262K · in $0.60 · out $3.60

Qwen3.5 Plus

Former Plus tier, superseded by Qwen3.6/3.7 Plus. 1M context, thinking, vision.

1M · in $0.40 · out $2.40

Qwen3 Coder Next

Budget agentic coder on the Qwen3-Next architecture. 256K context, non-thinking, tiered pricing.

262K · in $0.30 · out $1.50

DeepSeek V3.2 (Alibaba)

DeepSeek V3.2 served via Alibaba Model Studio (superseded by V4). Thinking.

131K · in $0.57 · out $1.71

Qwen3 VL Flash

Budget VL tier below Qwen3 VL Plus. 256K context, thinking, vision.

262K · in $0.05 · out $0.40

Qwen3 Vl 235b A22b Thinking

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 Vl 235b A22b Instruct

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 Coder Plus

Agentic coding model with very long context. Tiered pricing by input length (up to 1M).

1M · in $1.00 · out $5.00

Qwen3 Max

Retired 2025 flagship, superseded by Qwen3.6+ Max. 256K context; the base id now serves the thinking-capable 2026-01-23 snapshot.

262K · in $1.20 · out $6.00

Qwen3 Next 80b A3b Thinking

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 Next 80b A3b Instruct

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 30b A3b Thinking 2507

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 30b A3b Instruct 2507

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 Coder Flash

Former budget coder, superseded by Qwen3 Coder Next (which lacks its 1M context). Non-thinking.

1M · in $0.30 · out $1.50

Qwen3 235b A22b Thinking 2507

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 30b A3b

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 32b

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 14b

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 235b A22b

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 8b

Alibaba model (not yet curated).

131K · in - · out -

Qwen Plus

Balanced quality, speed, and cost with hybrid thinking. 1M context.

1M · in $0.40 · out $1.20

Qwen3 235b A22b Instruct 2507

Alibaba model (not yet curated).

131K · in - · out -

GLM-5.2 Fast (Alibaba)

Zhipu GLM-5.2 fast-serving tier via Alibaba Model Studio (preview). Same model, lower latency, ~2x price.

1M · in $2.80 · out $8.80

Qvq Max

Alibaba model (not yet curated).

131K · in - · out -

Qwen Coder Plus

Alibaba model (not yet curated).

131K · in - · out -

Qwen Flash

Fast and very low cost with hybrid thinking. 1M context.

1M · in $0.05 · out $0.40

Qwen Max

Best quality of the stable commercial line. 32K context.

33K · in $1.60 · out $6.40

Qwen Plus

Balanced quality, speed, and cost with hybrid thinking. 1M context.

1M · in $0.40 · out $1.20

Qwen Turbo

Fastest and cheapest for simple tasks. 1M context.

1M · in $0.05 · out $0.20

Qwen Vl Max

Alibaba model (not yet curated).

131K · in - · out -

Qwen Vl Plus

Alibaba model (not yet curated).

131K · in - · out -

Qwen3.7 Max

Flagship agent model with native extended thinking and 1M context. Text-only; strong at coding, productivity, and long-horizon autonomous tasks.

1M · in $2.50 · out $7.50

Qwen3 Coder 480b A35b Instruct

Alibaba model (not yet curated).

131K · in - · out -

Qwen3 Max

Retired 2025 flagship, superseded by Qwen3.6+ Max. 256K context; the base id now serves the thinking-capable 2026-01-23 snapshot.

262K · in $1.20 · out $6.00

Qwen3 VL Plus

Current vision-language model with strong visual reasoning and thinking. Tiered pricing by input length (up to 256K).

262K · in $0.20 · out $1.60

Qwq Plus

Alibaba model (not yet curated).

131K · in - · out -

59 models · sorted by release date · prices in USD per 1M tokens · refreshed hourlyCompare every model across vendors ->

Get started in 3 steps

1

Create an API key at the Alibaba console.

2

Paste it into Big-AGI's model settings.

3

Start chatting, or Beam it against other models and fuse the answers.

Running Alibaba in Big-AGI

Add your Alibaba Cloud Model Studio API key and run the Qwen family, plus the other models Alibaba serves, at Alibaba's own API rates. Big-AGI adds no markup and no intermediary: billing runs directly between you and Alibaba Cloud, and your keys stay in your browser.

Your key, your billing. Usage is billed by Alibaba Cloud to your account. Big-AGI does not meter or charge for model usage.
Chat models only. Alibaba's raw catalog mixes Qwen chat models in with OCR, speech, translation, and call-center products. Big-AGI filters those out so only genuine chat models show up, which Alibaba's own console does not do for you.
Third-party models, labeled honestly. Model Studio also hosts other labs' models at Alibaba's own pricing. Big-AGI tags them "(Alibaba)" so you never mistake a hosted copy for the vendor-direct version.

Why not just the Qwen app?

Alibaba's own Qwen app is free and covers general chat well, but it cannot put Qwen's answer next to GPT's, Claude's, or Gemini's on the same prompt. Beam can: agreement across labs is a stronger signal than any single model's confidence, and disagreement tells you where to dig further. You also get parameters the app hides (temperature, system prompt, per-turn model swaps), and a key that stays in your browser instead of sitting on Alibaba's servers.

Your keys and your data

Turn on Direct Connection and the browser calls Alibaba directly, bypassing the Big-AGI server, whenever your key is client-side and Alibaba allows it. Your keys stay in your browser. Chats are stored locally first and sync only if you turn it on. The AI Inspector opens on any message to show the exact request sent to Alibaba, the token counts, and a cost estimate for that call.

Qwen in Beam

Run Qwen in parallel with Claude, GPT, and Gemini on the same prompt, then reach for Fusions: several strategies that combine, cross-check, and synthesize the parallel answers, which beats just picking the single best one. Parallel runs use more tokens than a single chat.

Bring your Alibaba key. Keep control.

Your key, your data, your choice of model. Big-AGI's Open branch is open source and self-hostable, so you can check exactly how Alibaba is called.

<- All Models

Alibaba

Anthropic

AWS Bedrock

Azure

Cerebras

DeepSeek

Fireworks AI

Google Gemini

Groq

MiniMax

Mistral

Modular

Moonshot

NVIDIA

OpenAI

OpenRouter

Perplexity

Sakana AI

SpaceXAI

Together AI

Z.ai

BIG-AGI

Product

Features Models Controls What's new BEAM Technology

Resources

Documentation Discord GitHub

Company

Email Us Privacy Terms

© 2026 Token Fabrics·Built with passion in San Diego