BenchLeader
AlibabaReasoning modelNewReleased 16 Nov 2026Compare with another model

Qwen4 Max

Launch tracker · day 4

Qwen4 Max at launch: the first independent results

As of 19 Nov 2026, Qwen4 Max (Alibaba, released 16 Nov 2026) has independent results from Artificial Analysis (AA Intelligence Index v5.0.1 42.9), LMArena (Arena WebDev 1750), Vals AI (Vals Index 63.2), LiveBench (LiveBench 2.0 73.1%) and Design Arena (Design Arena Agentic 1356). It has no BenchLeader Index yet: the Index needs at least five independent results across two categories. It is listed at $1.50 in / $4.50 out per 1M tokens, with a 992k-token context window. Not yet measured by Epoch AI, ARC Prize, Scale AI and SWE-bench.

Released
16 Nov 2026 (day 4)
Provider
Alibaba
Price
$1.50 in / $4.50 out per 1M tokens
Context window
992k tokens
Open weights
No
BenchLeader Index
Not yet (needs five results in two categories)
Independent boardResults so far
Artificial AnalysisAA Intelligence Index v5.0.1 42.9 · AA-Omniscience v1.1 27.7 · HLE-Diamond (AA) 44.9% · more below
LMArenaArena WebDev 1750 · Arena Agent 7.3 · Arena Agent Task Completion 8.8
Vals AIVals Index 63.2 · Excel Modeling Benchmark v1.1 70.8% · Finance Agent v2.1 66.7% · more below
LiveBenchLiveBench 2.0 73.1% · LiveBench 2.0 Reasoning 92.2% · LiveBench 2.0 Coding 69.8% · more below
Design ArenaDesign Arena Agentic 1356

Boards usually post results within days of a launch; this page is rebuilt every three hours.

Scores from independent publishers only; benchmark figures a provider publishes about its own model are not shown.

Reasoning effort
Price#186 of 263
$2.25per 1M tokens
Dearer than most
2.5× the median
Speed#148 of 216
59tokens/s
Below the median
0.7× the median

Good value at this level: no ranked model within 2 Index points costs three quarters of the price or less.

No cheaper ranked model scores higher: it is on the price frontier.

As of 19 Nov 2026, Qwen4 Max by Alibaba is not yet ranked on the BenchLeader Index, which needs at least five independent results across two categories. API pricing: $1.50 per 1M input tokens and $4.50 per 1M output tokens ($2.25 blended at 3:1), list price on 19 Nov 2026. It streams about 59 tokens/s (Artificial Analysis). Released 16 Nov 2026.

Key facts (as of 19 Nov 2026)

  • Qwen4 Max API pricing: $1.50 per 1M input tokens and $4.50 per 1M output tokens ($2.25 blended at 3:1) (list price, 19 Nov 2026).
  • Qwen4 Max streams about 59 tokens/s (Artificial Analysis, 19 Nov 2026).
  • Qwen4 Max was released on 16 Nov 2026 (Alibaba).

In detail. Qwen4 Max is an Alibaba proprietary reasoning model, released 16 Nov 2026. It is not yet ranked: the index needs at least five independent results across two categories. At $2.25 per million tokens blended it is pricier than most ranked models. Output speed of 59 tokens per second puts it slower than most, with a first answer in 139.2 s. It has been measured at 2 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 16 Nov 2026.

Blended price
$2.25/M
$1.50 in · $4.50 out
via Artificial Analysis
Output speed
59 tok/s
measured by Artificial Analysis
First answer
139 s
first token 1.99 s
Context
992k
via LiteLLM price table
Accepts
Text · Images · Video
via models.dev, OpenRouter
Cost per task
$3.27
30 min a task, AA Index
Tokens per task
105k
66% of it reasoning
Released
16 Nov 2026
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Reasoning59
  2. Coding55
  3. Knowledge59

Get an email when Qwen4 Max gets its first benchmark results

And when Alibaba ships a new model, plus the weekly review on Mondays.

Choose providers and emails

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.

EffortIndexRankSpeedFirst answerChat reply costCategories
xhigh––59 tok/s139 s$0.0020Agents & tools 60 · Coding 75 · Composite 57
not statedbest––59 tok/s139 s$0.0020Coding 55 · Knowledge 59 · Reasoning 59

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmarkxhighnot statedSourceTrend
LiveBench 2.0 Reasoningnot in index92.2%#4–LiveBench
HLE-Diamond (AA)effort not stated–44.9%#47Artificial Analysis
CritPt v2 hard subset (AA)effort not stated–43.0%#71Artificial Analysis

Coding

Benchmarkxhighnot statedSourceTrend
Arena WebDev1750#9–Arena (LMArena)
LiveBench 2.0 Codingnot in index69.8%#25–LiveBench
Vibe Code Bench v1.1not in index86.7%#18–Vals AI
Terminal-Bench 4.0 (AA)not in indexeffort not stated–48.5%#46Artificial Analysis
FrontierCode 2.0 (AA)effort not stated–26.4%#62Artificial Analysis

Agents & tools

Benchmarkxhighnot statedSourceTrend
Arena Agent7.3#18–Arena (LMArena)
LiveBench Long Agenticnot in index46.6%#16–LiveBench
LiveBench Browsernot in index80.7%#12–LiveBench
GDPval-AA v2.1not in indexeffort not stated–1809#7Artificial Analysis
Analyst Agent (AA)not in indexeffort not stated–46.3%#25Artificial Analysis
APEX-Agents (AA)not in indexeffort not stated–59.1%#1Artificial Analysis
Finance Agent v2.1not in index66.7%#9–Vals AI
Harvey's Legal Agent Benchmark v1.1not in index21.3%#7–Vals AI
Legal Research Benchnot in index41.4%#35–Vals AI
Tax Agent Benchnot in index40.4%#14–Vals AI
Terminal-Bench 4.0 (Vals)not in index48.5%#18–Vals AI
Arena Agent Task Completionnot in index8.8#16–Arena (LMArena)
AutomationBenchnot in indexeffort not stated–72.6%#18Artificial Analysis
GDP.pdf (AA)not in indexeffort not stated–26.2%#50Artificial Analysis
Harvey LAB-AA v1.1not in indexeffort not stated–9.4%#8Harvey
EnterpriseOps-Gymnot in indexeffort not stated–48.5%#5Artificial Analysis
AA-Briefcase v1.1not in indexeffort not stated–1686#9Artificial Analysis
Design Arena Agenticnot in indexeffort not stated–1356#5Design Arena
Terminal-Bench-Science 0.1 (AA)not in indexeffort not stated–23.3%#58Artificial Analysis

Knowledge

Benchmarkxhighnot statedSourceTrend
LiveBench 2.0 Data Analysisnot in index83.4%#5–LiveBench
LiveBench Researchnot in index84.7%#17–LiveBench
AA-Omniscience v1.1effort not stated–27.7#57Artificial Analysis
Excel Modeling Benchmark v1.1not in index70.8%#21–Vals AI
AA-Omniscience v1.1: accuracynot in indexeffort not stated–43.7%#72Artificial Analysis
AA-Omniscience v1.1: non-hallucinationnot in indexeffort not stated–71.6%#39Artificial Analysis

Multimodal

BenchmarkxhighSourceTrend
LiveBench Designnot in index79.1%#14LiveBench

Long context

Benchmarknot statedSourceTrend
MLCRnot in index19.4%#75Artificial Analysis

Composite

Benchmarkxhighnot statedSourceTrend
LiveBench 2.073.1%#14–LiveBench
AA Intelligence Index v5.0.1not in indexeffort not stated–42.9#52Artificial Analysis
Vals Indexnot in index63.2#14–Vals AI
Vals Finance Indexnot in index62.0%#9–Vals AI
Vals Legal Indexnot in index70.7%#18–Vals AI

API pricing: where to run it

Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.

ProviderSpeedFirst tokenInput $/MOutput $/MContextQuantisation
Alibaba Cloud Int.49 tok/s2.23 s$1.43$4.281M–

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.071 per 1M. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0020$0.00152.4 min
Summarize a 30-page report12,000 / 600$0.021$0.00782.5 min
Code edit6,000 / 1,500$0.016$0.00932.7 min
Agentic coding session60,000 / 4,000$0.108$0.0443.5 min
Structured extraction2,000 / 200$0.0039$0.00182.4 min

More from Alibaba

New this month

Compare Qwen4 Max with…

See also

Questions people ask

Why does Qwen4 Max have no BenchLeader Index score?
The index needs at least five independent results across at least two categories, and as of 19 Nov 2026 Qwen4 Max does not have them yet. Launch-post scores are never used, so it is ranked once independent publishers have measured it.
How much does Qwen4 Max cost per 1M tokens?
As of 19 Nov 2026, Qwen4 Max costs $1.50 per 1M input tokens and $4.50 per 1M output tokens, $2.25 blended (three input tokens per output token).
Is Qwen4 Max better than Qwen3.8 Max?
As of 19 Nov 2026, only one of these has enough independent results to rank; on price, Qwen4 Max is 1.3× cheaper. The head-to-head page has the benchmark-by-benchmark detail.
When was Qwen4 Max released?
Alibaba released Qwen4 Max on 16 Nov 2026. BenchLeader has results for it at 2 reasoning-effort settings.

Data as of 19 Nov 2026. Compare these configurations.

Cite as: BenchLeader, “Qwen4 Max: benchmarks, pricing, speed and rank”, data as of 19 Nov 2026, https://www.benchleader.com/models/qwen4.