BenchLeader

Qwen3 235B A22B

Reasoning effort
Quality#174 of 339
49.5Index
Below the median
−0.4 on the median
Price#149 of 255
$1.23per 1M tokens
Dearer than most
1.4× the median
Value#155 of 255
40.4points per $
Below the median
0.8× the median

Better value: Qwen3.5-4B scores 49.3 for $0.060/M, 20× cheaper.

72 cheaper ranked models score higher: Kimi K2.5 (59.4, $1.20), Qwen3.6 Plus (59.0, $1.13), Gemini 3 Flash (58.9, $1.13) and 69 more.

As of 15 Nov 2026, Qwen3 235B A22B by Alibaba scores 49.5 on the BenchLeader Index (611 ranked configurations), with its best results in maths (59) and weakest in instruction following (43). API pricing: $0.700 per 1M input tokens and $2.80 per 1M output tokens ($1.23 blended at 3:1), list price on 15 Nov 2026. Released 28 Apr 2025.

Key facts (as of 15 Nov 2026)

  • Qwen3 235B A22B API pricing: $0.700 per 1M input tokens and $2.80 per 1M output tokens ($1.23 blended at 3:1) (list price, 15 Nov 2026).
  • Qwen3 235B A22B was released on 28 Apr 2025 (Alibaba).

In detail. Qwen3 235B A22B is an Alibaba open-weights reasoning model, released 28 Apr 2025. Its best configuration ranks #359 of 611 on the BenchLeader Index at 49.5 ±4.0, in the lower half. The ± is the point: 265 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in maths (59) and lowest in instruction following (43). At $1.23 per million tokens blended it is mid-priced. It has been superseded by Qwen3 235B A22B 2507, which is the version to compare against today. It has been measured at 3 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 11 Nov 2026.

Blended price
$1.23/M
$0.700 in · $2.80 out
via models.dev
Output speed
–
First answer
–
Context
131k
via models.dev
Cost per task
–
Tokens per task
–
Released
28 Apr 2025
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index50
  2. Reasoning45
  3. Coding46
  4. Maths59
  5. Knowledge50
  6. Instruction following43
  7. Human preference54

Get an email when Alibaba ships a new model

And once when it gets ranked, plus the weekly review on Mondays.

Choose providers and emails

Versions

Alibaba has shipped 2 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.

ModelReleasedIndexRank
Qwen3 235B A22B 2507thinking22 Jul 202551.8#301
Qwen3 235B A22Bthis page28 Apr 202549.5#359

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
no reasoning47.4#409––$0.0011Agents & tools 39 · Coding 57 · Human preference 57 · Instruction following 40 · Long context 23 · Maths 56 · Reasoning 54
thinking––––$0.0011Agents & tools 39 · Instruction following 42 · Long context 23 · Reasoning 63
not statedbest49.5#359––$0.0011Coding 46 · Human preference 54 · Instruction following 43 · Knowledge 50 · Maths 59 · Reasoning 45

How its index has moved

8 Sep 2026 to 15 Nov 2026
394756
  • not stated
  • no reasoning

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

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

Benchmarkno reasoningthinkingnot statedSourceTrend
GPQA Diamondeffort not stated––70.7%#182Epoch AI Benchmarking Hub
SimpleBencheffort not stated––31.0%#88SimpleBench
Arena Hard Prompts1420#155–1392#186Arena (LMArena)
GPQA Diamond (AA)not in index61.3%#37470.0%#304–Artificial Analysis
Humanity's Last Exam (AA)not in index4.2%#55011.0%#324–Artificial Analysis
GPQA Diamond (Vals)not in indexeffort not stated––70.2%#91Vals AI
Kagi LLM Benchmark55.0%#6669.4%#29–Kagi LLM Benchmark
GPQA Diamond (HELM)not in indexeffort not stated––62.3%#24HELM Capabilities
CritPtnot in index0.0%#3370.0%#337–Artificial Analysis
SciPredictnot in indexeffort not stated––16.6%#13Scale AI SEAL
LMCAnot in indexeffort not stated––25.0%#157Epoch AI Benchmarking Hub
DTBenchnot in indexeffort not stated––75.7%#155Epoch AI Benchmarking Hub
ForecastBenchnot in indexeffort not stated––63.9%#35ForecastBench

Coding

Benchmarkno reasoningthinkingnot statedSourceTrend
WeirdMLeffort not stated––37.3%#119WeirdML
Arena Coding1444#158–1432#180Arena (LMArena)
LiveCodeBencheffort not stated––70.6%#86Vals AI
SWE-Bench Proeffort not stated––21.4%#18Scale AI SEAL
Aider Polyglot59.6%#23––Aider polyglot leaderboard
SWE-Bench Pro (private)not in indexeffort not stated––21.4%#18Scale AI SEAL
Design Arena Websitenot in indexeffort not stated––1022#165Design Arena
LiveCodeBench (AA)not in index34.3%#18862.2%#104–Artificial Analysis

Agents & tools

Benchmarkno reasoningthinkingSourceTrend
Terminal-Bench Hard6.1%#2716.1%#271Artificial Analysis
τ²-Bench Telecom (AA)not in index27.2%#28224.0%#308Artificial Analysis

Maths

Benchmarkno reasoningthinkingnot statedSourceTrend
MATH Level 5effort not stated––68.9%#41Epoch AI Benchmarking Hub
AIME (Vals)effort not stated––84.0%#40Vals AI
MGSMeffort not stated––92.5%#17Vals AI
Omni-MATH (HELM)effort not stated––54.8%#15HELM Capabilities
MATH 500not in indexeffort not stated––94.6%#8Vals AI
Arena Maths1390#166–1391#165Arena (LMArena)
AIME 2025 (AA)not in index23.7%#19282.0%#58–Artificial Analysis

Knowledge

Benchmarknot statedSourceTrend
MMLU-Pro81.3%#81Vals AI
LegalBench80.2%#89Vals AI
TaxEval70.7%#88Vals AI
MedQA90.6%#44Vals AI
MultiNRC17.6%#38Scale AI SEAL
MMLU-Pro (HELM)not in index81.7%#15HELM Capabilities

Instruction following

Benchmarkno reasoningthinkingnot statedSourceTrend
IFBench36.6%#30438.7%#282–Artificial Analysis
MultiChallengeeffort not stated––41.2%#33Scale AI SEAL
IFEval (HELM)effort not stated––81.6%#40HELM Capabilities
WildBench (HELM)effort not stated––82.8%#20HELM Capabilities
Arena Instruction Followingnot in index1381#170–1357#192Arena (LMArena)

Human preference

Benchmarkno reasoningthinkingnot statedSourceTrend
Arena Text1402#155–1375#185Arena (LMArena)
EQ-Bench Creative Writing v3not in index–1366#78–EQ-Bench
Arena Creative Writingnot in index1366#157–1323#202Arena (LMArena)
Arena Multi-turnnot in index1409#151–1369#188Arena (LMArena)
Arena Longer Queriesnot in index1410#153–1378#188Arena (LMArena)
Lech Mazur Creative Writingnot in indexeffort not stated––83.0%#9Lech Mazur

Long context

Benchmarkno reasoningthinkingSourceTrend
AA-LCR v1.10.0%#5170.0%#517Artificial Analysis

Composite

Benchmarkno reasoningthinkingnot statedSourceTrend
Epoch Capabilities Indexnot in indexeffort not stated––138.4#120Epoch AI Benchmarking Hub
AA Intelligence Index v4.3.2not in index8.3#4519.5#407–Artificial Analysis
AA Progress Index v1not in index175.4#492208.9#385–Artificial Analysis
HELM Capabilities meannot in indexeffort not stated––72.6%#19HELM Capabilities

Safety & honesty

Benchmarknot statedSourceTrend
FORTRESSnot in index61.4%#4Scale AI SEAL
MASKnot in indexpossible contamination56.4%#43Scale AI SEAL

Price history

The price above, per 1M tokens, every day it changed.

$0.00$0.770$1.54$2.31$3.08Sep 26Oct 26Nov 26
input output
Now $0.700 in · $2.80 out

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. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0011––
Summarize a 30-page report12,000 / 600$0.010––
Code edit6,000 / 1,500$0.0084––
Agentic coding session60,000 / 4,000$0.053––
Structured extraction2,000 / 200$0.0020––

More from Alibaba

New this month

See also

Embed or cite

Cite this

The ranking changes daily, so a reference carries the data date. APA style:

BenchLeader. (2026, November 15). Qwen3 235B A22B on the BenchLeader Index (method v1.4) [Data set]. Retrieved November 15, 2026, from https://www.benchleader.com/models/qwen3-235b-a22b

BibTeX:

@misc{benchleader_qwen3_235b_a22b_20261115,
  author       = {{BenchLeader}},
  title        = {{Qwen3 235B A22B on the BenchLeader Index (method v1.4)}},
  year         = {2026},
  month        = nov,
  howpublished = {\url{https://www.benchleader.com/models/qwen3-235b-a22b}},
  note         = {Data as of 2026-11-15}
}

Every badge style, the embeddable top-10 widget and how to cite the Index

Questions people ask

What are Qwen3 235B A22B's benchmark scores?
As of 15 Nov 2026, Qwen3 235B A22B scores 49.5 ±4.0 on the BenchLeader Index, strongest in maths (59) and weakest in instruction following (43). That rests on independent benchmark results only, and every score on this page links to its publisher.
How does Qwen3 235B A22B rank?
Qwen3 235B A22B is #359 of 611 ranked configurations on 15 Nov 2026. 265 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Fable 5.2 (max effort) at 72.7.
How much does Qwen3 235B A22B cost per 1M tokens?
As of 15 Nov 2026, Qwen3 235B A22B costs $0.700 per 1M input tokens and $2.80 per 1M output tokens, $1.23 blended (three input tokens per output token). The list price last changed on 4 Nov 2026.
When was Qwen3 235B A22B released?
Alibaba released Qwen3 235B A22B on 28 Apr 2025. BenchLeader has results for it at 3 reasoning-effort settings.

Data as of 15 Nov 2026. Compare these configurations.

Cite as: BenchLeader, “Qwen3 235B A22B: benchmarks, pricing, speed and rank”, data as of 15 Nov 2026, https://www.benchleader.com/models/qwen3-235b-a22b.