upcube
Alibaba•Open weights•November 3, 2025

Qwen3 Max Thinking (Preview)

Intelligence, Performance & Price Analysis

Source:Artificial Analysis scrapeType:Open weights
Intelligence
25

score

Qwen3 Max Thinking (Preview) scores 25 (estimated) on the Artificial Analysis Intelligence Index, …

vs class avg: 28

Speed
59.0

output tokens/sec

Qwen3 Max Thinking (Preview) generates output at 59.0 tokens per second (based on Alibaba's API), …

Latency
4.12s

TTFT

Qwen3 Max Thinking (Preview) has a time to first token (TTFT) of 4. …

Input Price#52/169
$1.20

/ 1M tokens

Qwen3 Max Thinking (Preview) costs $1.20 per 1M input tokens (better than average, median: $1. …

Output Price#60/169
$6.00

/ 1M tokens

Qwen3 Max Thinking (Preview) costs $1.20 per 1M input tokens (better than average, median: $1. …

Decision Overview

Ethen's intelligence-driven routing assessment for this model

Overall verdict

Strong general-purpose model

Suitable for most production tasks, but high-volume or repetitive work should still be compared against cheaper routes.

Best for

  • Code generation
  • Content analysis
  • Structured extraction
  • Customer-facing chat
  • High-value tasks where quality outweighs cost

Weak or unsuitable

  • Extremely long-context tasks
  • Ultra-low-latency requirements
  • Routine high-volume generation
  • Simple extraction or labeling

Routing recommendation

1

Lower-cost models for repetitive or high-volume work

2

Cache-backed reuse for repeated prompts

Cost pressure

Premium pricing — $1.20 / M input, $6.00 / M output. Use cheaper routes for repetitive or low-risk work.

Model-fit matrix

Use this model where capability pays for itself; route away where volume, latency, or budget dominate.

Use caseFitReasonRouting note
Complex reasoning & agentic workflowsStrongIntelligence score 25 supports capable reasoning, but very hard tasks may benefit from higher-tier models.Good for most complex tasks; consider a frontier model for the hardest 10%.
High-volume chat & customer-facingStrongOutput speed 59.0 tokens/sec is adequate for chat.Suitable for chat; monitor latency under concurrent load.
Latency-sensitive applicationsModerateTTFT 4.12s — latency may be noticeable in interactive use.Consider routing latency-critical paths to faster models.
Cost-sensitive pipelinesModerateOutput pricing at $6.00 is premium. Route high-volume simple tasks to cheaper alternatives.Use only for high-value tasks; route simple queries to budget models.

Cost pressure

High

Pricing is premium — input $1.20, output $6.00. This model is expensive for high-volume or output-heavy workloads.

Cheaper substitutes

  • Lower-cost models for repetitive or high-volume work
  • Cache-backed reuse for repeated prompts

Route away when

  • Output-heavy tasks
  • High-frequency chat at scale
  • Simple queries that cheaper models handle well

Highlights

Key benchmark results at a glance.

Intelligence

Intelligence position

Open section

Artificial Analysis Intelligence Index · Higher is better · Evaluation results measured independently by Artificial Analysis

Qwen3 Max Thinking (Preview) scores 25 (estimated) on the Artificial Analysis Intelligence Index, placing it below average among other reasoning models in a similar price tier (median: 28).

0204060Claude Fable 5 (with fallback): 6060AClaude Fable 5 (with fal…GPT-5.5 (xhigh): 5555AIGPT-5.5 (xhigh)GLM-5.2 (max): 5151?GLM-5.2 (max)Gemini 3.5 Flash: 5050GGemini 3.5 FlashMiniMax-M3: 4444?MiniMax-M3DeepSeek V4 Pro (max): 4444DDeepSeek V4 Pro (max)Kimi K2.6: 4444KKimi K2.6Muse Spark: 4343?Muse SparkNemotron 3 Ultra: 3838NNemotron 3 UltraGrok 4.3 (high): 3838xGrok 4.3 (high)Qwen3 Max Thinking (Preview): 2525QQwen3 Max Thinking (Prev…gpt-oss-120b (high): 2424AIgpt-oss-120b (high)

Speed

Speed position

Open section

Output tokens per second · Higher is better · Evaluation results measured independently by Artificial Analysis

Qwen3 Max Thinking (Preview) generates output at 59.0 tokens per second (based on Alibaba's API), which is below average compared to other reasoning models in a similar price tier (median: 81.2 t/s).

088175263350gpt-oss-120b (high): 314314AIgpt-oss-120b (high)Nemotron 3 Ultra: 250250NNemotron 3 UltraGLM-5.2 (max): 218218?GLM-5.2 (max)Qwen3.7 Max: 206206QQwen3.7 MaxGemini 3.5 Flash: 192192GGemini 3.5 FlashGPT-5.4 mini (xhigh): 179179AIGPT-5.4 mini (xhigh)GPT-5.4 (xhigh): 175175AIGPT-5.4 (xhigh)Grok 4.3 (high): 164164xGrok 4.3 (high)Gemini 3.1 Pro Preview: 152152GGemini 3.1 Pro PreviewMiniMax-M3: 9999?MiniMax-M3Claude Sonnet 5 (max): 8989AClaude Sonnet 5 (max)GPT-5.5 (xhigh): 8888AIGPT-5.5 (xhigh)

Context, Cost & Pricing

Price/value position

Open section

Weighted average cost (USD) per Intelligence Index task · Lower is better · Evaluation results measured independently by Artificial Analysis

Pricing is premium — input $1.20, output $6.00. This model is expensive for high-volume or output-heavy workloads.

$0.00$1.00$2.00$3.00$4.00$5.00DeepSeek V4 Pro (max): $0.04$0.04DDeepSeek V4 Pro (max)gpt-oss-120b (high): $0.06$0.06AIgpt-oss-120b (high)MiniMax-M3: $0.12$0.12?MiniMax-M3Grok 4.3 (high): $0.14$0.14xGrok 4.3 (high)Nemotron 3 Ultra: $0.24$0.24NNemotron 3 UltraKimi K2.6: $0.35$0.35KKimi K2.6GLM-5.2 (max): $0.37$0.37?GLM-5.2 (max)Gemini 3.5 Flash: $0.59$0.59GGemini 3.5 FlashGPT-5.5 (xhigh): $0.86$0.86AIGPT-5.5 (xhigh)Claude Fable 5 (with fallback): $2.75$2.75AClaude Fable 5 (with fal…

Intelligence & Quality

Benchmark scores and quality indices measuring model capability.

AA-Omniscience Index

AA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct. · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

0204060Claude Fable 5 (with fallback): 4040AClaude Fable 5 (with fal…Gemini 3.1 Pro Preview: 3333GGemini 3.1 Pro PreviewClaude Opus 4.8 (max): 2727AClaude Opus 4.8 (max)Claude Opus 4.7 (max): 2626AClaude Opus 4.7 (max)Gemini 3.5 Flash: 2323GGemini 3.5 FlashGPT-5.5 (xhigh): 2020AIGPT-5.5 (xhigh)GPT-5.5 (high): 1818AIGPT-5.5 (high)Grok 4.3 (high): 1818xGrok 4.3 (high)GPT-5.5 (medium): 1717AIGPT-5.5 (medium)Claude Sonnet 5 (max): 1515AClaude Sonnet 5 (max)

Artificial Analysis Intelligence Index by Open Weights / Proprietary

Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

0204060Claude Fable 5 (with fallback): 6060AClaude Fable 5 (with fal…Claude Opus 4.8 (max): 5656AClaude Opus 4.8 (max)GPT-5.5 (xhigh): 5555AIGPT-5.5 (xhigh)Claude Opus 4.7 (max): 5454AClaude Opus 4.7 (max)Claude Sonnet 5 (max): 5353AClaude Sonnet 5 (max)GPT-5.5 (high): 5353AIGPT-5.5 (high)GPT-5.4 (xhigh): 5151AIGPT-5.4 (xhigh)GLM-5.2 (max): 5151?GLM-5.2 (max)GPT-5.5 (medium): 5050AIGPT-5.5 (medium)Gemini 3.5 Flash: 5050GGemini 3.5 Flash

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

0204060Claude Fable 5 (with fallback): 6060AClaude Fable 5 (with fal…Claude Opus 4.8 (max): 5656AClaude Opus 4.8 (max)GPT-5.5 (xhigh): 5555AIGPT-5.5 (xhigh)Claude Opus 4.7 (max): 5454AClaude Opus 4.7 (max)Claude Sonnet 5 (max): 5353AClaude Sonnet 5 (max)GPT-5.5 (high): 5353AIGPT-5.5 (high)GPT-5.4 (xhigh): 5151AIGPT-5.4 (xhigh)GLM-5.2 (max): 5151?GLM-5.2 (max)GPT-5.5 (medium): 5050AIGPT-5.5 (medium)Gemini 3.5 Flash: 5050GGemini 3.5 Flash

Intelligence

Artificial Analysis Intelligence Index · Higher is better · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

0204060Claude Fable 5 (with fallback): 6060AClaude Fable 5 (with fal…GPT-5.5 (xhigh): 5555AIGPT-5.5 (xhigh)GLM-5.2 (max): 5151?GLM-5.2 (max)Gemini 3.5 Flash: 5050GGemini 3.5 FlashMiniMax-M3: 4444?MiniMax-M3DeepSeek V4 Pro (max): 4444DDeepSeek V4 Pro (max)Kimi K2.6: 4444KKimi K2.6Muse Spark: 4343?Muse SparkNemotron 3 Ultra: 3838NNemotron 3 UltraGrok 4.3 (high): 3838xGrok 4.3 (high)Qwen3 Max Thinking (Preview): 2525QQwen3 Max Thinking (Prev…

Speed & Throughput

Output tokens per second and generation throughput.

Output Speed

Output tokens per second · Higher is better · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

088175263350gpt-oss-120b (high): 314314AIgpt-oss-120b (high)Nemotron 3 Ultra: 250250NNemotron 3 UltraGLM-5.2 (max): 218218?GLM-5.2 (max)Qwen3.7 Max: 206206QQwen3.7 MaxGemini 3.5 Flash: 192192GGemini 3.5 FlashGPT-5.4 mini (xhigh): 179179AIGPT-5.4 mini (xhigh)GPT-5.4 (xhigh): 175175AIGPT-5.4 (xhigh)Grok 4.3 (high): 164164xGrok 4.3 (high)Gemini 3.1 Pro Preview: 152152GGemini 3.1 Pro PreviewMiniMax-M3: 9999?MiniMax-M3

Speed

Output tokens per second · Higher is better · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

088175263350gpt-oss-120b (high): 314314AIgpt-oss-120b (high)Nemotron 3 Ultra: 250250NNemotron 3 UltraGLM-5.2 (max): 218218?GLM-5.2 (max)Gemini 3.5 Flash: 192192GGemini 3.5 FlashGrok 4.3 (high): 164164xGrok 4.3 (high)MiniMax-M3: 9999?MiniMax-M3GPT-5.5 (xhigh): 8888AIGPT-5.5 (xhigh)Kimi K2.6: 7676KKimi K2.6DeepSeek V4 Pro (max): 7272DDeepSeek V4 Pro (max)Claude Fable 5 (with fallback): 7171AClaude Fable 5 (with fal…Qwen3 Max Thinking (Preview): 5959QQwen3 Max Thinking (Prev…

Latency & Response Time

Time to first token and end-to-end response latency.

End-to-End Response Time

Seconds to output 500 tokens, including reasoning model 'thinking' time · Lower is better · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

0s2.5s5s7.5s10sKimi K2.7 Code: 9.7s9.7sKKimi K2.7 CodeClaude Opus 4.7 (max): 8.9s8.9sAClaude Opus 4.7 (max)Qwen3 Max Thinking (Preview): 8.5s8.5sQQwen3 Max Thinking (Prev…Claude Opus 4.8 (max): 7.6s7.6sAClaude Opus 4.8 (max)Grok Build 0.1 0616: 7.2s7.2sxGrok Build 0.1 0616DeepSeek V4 Pro (max): 7s7sDDeepSeek V4 Pro (max)GPT-5.5 (medium): 6.9s6.9sAIGPT-5.5 (medium)Kimi K2.6: 6.6s6.6sKKimi K2.6GLM-5.1: 6.3s6.3s?GLM-5.1GPT-5.5 (low): 6.1s6.1sAIGPT-5.5 (low)

Latency: Time To First Answer Token

Seconds to first answer token received · Accounts for reasoning model 'thinking' time · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

0s25s50s75s100sDeepSeek V4 Pro (max): 61s61sDDeepSeek V4 Pro (max)Kimi K2.6: 58s58sKKimi K2.6GLM-5.1: 48s48s?GLM-5.1Kimi K2.7 Code: 43s43sKKimi K2.7 CodeQwen3 Max Thinking (Preview): 34s34sQQwen3 Max Thinking (Prev…Grok Build 0.1 0616: 29s29sxGrok Build 0.1 0616MiniMax-M3: 20s20s?MiniMax-M3Qwen3.7 Max: 12s12sQQwen3.7 MaxGLM-5.2 (max): 9.2s9.2s?GLM-5.2 (max)Nemotron 3 Ultra: 9.1s9.1sNNemotron 3 Ultra

Context, Cost & Pricing

Context window size, cost per task, and token pricing.

Context Window

Context window: tokens limit · Higher is better · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

0500k1M1.5M2MGPT-5.4 (xhigh): 1.1M1.1MAIGPT-5.4 (xhigh)Gemini 3.5 Flash: 1M1MGGemini 3.5 FlashClaude Fable 5 (with fallback): 1M1MAClaude Fable 5 (with fal…GLM-5.2 (max): 1M1M?GLM-5.2 (max)DeepSeek V4 Pro (max): 1M1MDDeepSeek V4 Pro (max)Grok 4.3 (high): 1M1MxGrok 4.3 (high)MiniMax-M3: 1M1M?MiniMax-M3Claude Opus 4.8 (max): 1M1MAClaude Opus 4.8 (max)Claude Opus 4.7 (max): 1M1MAClaude Opus 4.7 (max)Claude Sonnet 5 (max): 1M1MAClaude Sonnet 5 (max)Qwen3 Max Thinking (Preview): 262k262kQQwen3 Max Thinking (Prev…

Cost per Intelligence Index Task

Weighted average cost (USD) per Artificial Analysis Intelligence Index task, segmented by token type. Lower is better · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

00.51GPT-5.5 (xhigh): 0.10.1AIGPT-5.5 (xhigh)Claude Sonnet 4.6 (max): 0.10.1AClaude Sonnet 4.6 (max)GPT-5.5 (high): 0.10.1AIGPT-5.5 (high)Gemini 3.5 Flash: 0.10.1GGemini 3.5 FlashGPT-5.5 (medium): 0.10.1AIGPT-5.5 (medium)GPT-5.4 (xhigh): 0.10.1AIGPT-5.4 (xhigh)GPT-5.5 (low): 00AIGPT-5.5 (low)GLM-5.1: 00?GLM-5.1Qwen3.7 Max: 00QQwen3.7 MaxGemini 3.1 Pro Preview: 00GGemini 3.1 Pro Preview

Cost per Task

Weighted average cost (USD) per Intelligence Index task · Lower is better · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

$0.00$1.00$2.00$3.00$4.00$5.00Claude Fable 5 (with fallback): $2.75$2.75AClaude Fable 5 (with fal…GPT-5.5 (xhigh): $0.86$0.86AIGPT-5.5 (xhigh)Gemini 3.5 Flash: $0.59$0.59GGemini 3.5 FlashGLM-5.2 (max): $0.37$0.37?GLM-5.2 (max)Kimi K2.6: $0.35$0.35KKimi K2.6Nemotron 3 Ultra: $0.24$0.24NNemotron 3 UltraGrok 4.3 (high): $0.14$0.14xGrok 4.3 (high)MiniMax-M3: $0.12$0.12?MiniMax-M3gpt-oss-120b (high): $0.06$0.06AIgpt-oss-120b (high)DeepSeek V4 Pro (max): $0.04$0.04DDeepSeek V4 Pro (max)

Cost to Run Artificial Analysis Intelligence Index

Cost (USD) to run all evaluations in the Artificial Analysis Intelligence Index · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

050100150Claude Sonnet 4.6 (max): 149149AClaude Sonnet 4.6 (max)GPT-5.5 (xhigh): 132132AIGPT-5.5 (xhigh)GPT-5.5 (high): 119119AIGPT-5.5 (high)Gemini 3.5 Flash: 114114GGemini 3.5 FlashGPT-5.5 (medium): 9999AIGPT-5.5 (medium)GPT-5.4 (xhigh): 8989AIGPT-5.4 (xhigh)GPT-5.5 (low): 6565AIGPT-5.5 (low)Gemini 3.1 Pro Preview: 4949GGemini 3.1 Pro PreviewQwen3.7 Max: 4848QQwen3.7 MaxGLM-5.1: 4040?GLM-5.1

Pricing: Cache Hit, Input, and Output

Price (USD per M Tokens) · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

0250k500k750k1MClaude Opus 4.8 (max): 3131AClaude Opus 4.8 (max)Claude Sonnet 4.6 (max): 1818AClaude Sonnet 4.6 (max)GPT-5.4 (xhigh): 1818AIGPT-5.4 (xhigh)Gemini 3.1 Pro Preview: 1414GGemini 3.1 Pro PreviewClaude Sonnet 5 (max): 1212AClaude Sonnet 5 (max)Claude Sonnet 5 (Non-reasoning): 1212AClaude Sonnet 5 (Non-rea…Gemini 3.5 Flash: 1111GGemini 3.5 FlashQwen3.7 Max: 1010QQwen3.7 MaxQwen3 Max Thinking (Preview): 7.27.2QQwen3 Max Thinking (Prev…GLM-5.2 (max): 6.16.1?GLM-5.2 (max)cache Hitinputoutput

Time per Intelligence Index Task

Weighted average decode time (minutes) per task; excludes TTFT and overhead time · Lower is better · Evaluation results measured independently by Artificial Analysis

Source and methodology are summarized below

Chart source and provenance are listed in Methodology & sources below.

02468DeepSeek V4 Pro (max): 6.86.8DDeepSeek V4 Pro (max)Claude Opus 4.8 (max): 6.86.8AClaude Opus 4.8 (max)Claude Opus 4.7 (max): 5.95.9AClaude Opus 4.7 (max)Kimi K2.7 Code: 55KKimi K2.7 CodeGLM-5.1: 4.94.9?GLM-5.1Claude Fable 5 (with fallback): 4.84.8AClaude Fable 5 (with fal…MiniMax-M3: 3.93.9?MiniMax-M3Grok Build 0.1 0616: 3.83.8xGrok Build 0.1 0616GPT-5.5 (xhigh): 33AIGPT-5.5 (xhigh)GPT-5.4 (xhigh): 33AIGPT-5.4 (xhigh)

Specifications

Technical Specifications

Qwen3 Max Thinking (Preview) scores 25 (estimated) on the Artificial Analysis Intelligence Index, placing it below average among other reasoning models in a similar price tier (median: 28). Qwen3 Max Thinking (Preview) generates output at 59.0 tokens per second (based on Alibaba's API), which is below average compared to other reasoning models in a similar price tier (median: 81.2 t/s). Qwen3 Max Thinking (Preview) costs $1.20 per 1M input tokens (better than average, median: $1.50) and $6.00 per 1M output tokens (better than average, median: $8.40), based on Alibaba's API.

Qwen3 Max Thinking (Preview)

Model type
Open weights
Reasoning
Yes
Input modalities
Qwen3 Max Thinking (Preview) supports text input.
Output modalities
Qwen3 Max Thinking (Preview) supports text output.
Context window
260k tokens
Open weights / source
No, Qwen3 Max Thinking (Preview) is proprietary. The model weights are not publicly available.
Parameters
Qwen3 Max Thinking (Preview) is a proprietary model and Alibaba has not disclosed the model size or parameter count.
API availability
Yes, Qwen3 Max Thinking (Preview) is available via API through 1 provider.

Methodology & Provenance

This page is rendered from the normalized profile and page JSON for Qwen3 Max Thinking (Preview).

Benchmark values are preserved as normalized; only layout, disclosure ordering, and typography are adjusted for readability.

Frequently Asked Questions