Models / Qwen3.5-397B-A17B

Alibaba

Qwen3.5-397B-A17B

Qwen3.5-397B-A17B is strongest at reasoning and knowledge; worth a look for simple, high-volume work where cost matters most. The cheapest model of similar intelligence is DeepSeek-V4-Flash (0423), at $0.15 per million tokens against $1.29.

Results 16 Nov 2026 · 42 results · 5 sources

Test Qwen3.5-397B-A17B on your prompt → It runs beside the models you could use instead, with every answer checked and costed.

Price per million tokens
$0.55 in · $3.50 out
Typical host on OpenRouter, 10 Nov 2026 · compare prices
Speed
56 tokens a second; on its test prompt the answer starts after 1.5 s, thinking time included
Artificial Analysis, on its usual API, at no reasoning; 55–56 across 2 settings
Intelligence Index
21.4
Artificial Analysis, at no reasoning; 18.4–21.4 across 2 settings
Spring Prompt overall
Not ranked yet
needs our benchmarks and two groups of results
Developer
Alibaba, based in China
Weights published
Released
13 Feb 2026
262,144 tokens of context

Where it stands among the models you could choose

Available models grouped by price, intelligence, speed and developer location, plus models with published weights. Retired models are excluded.

AmongIntelligence IndexSpring Prompt overallPrice, blendedSpeed
Models available today 94th of 185Claude Opus 5.5 – $1.29 blended, 61st percentilecheapest: Mistral Nemo 54th of 65Nemotron 3.5 Super VL
Models priced $1–3 per million tokens 27th of 34Muse Spark 1.3 – sets this group 7th of 9Muse Spark 1.3
Models with similar intelligence sets this group – $1.29 blended, 66th percentilecheapest: DeepSeek-V4-Flash (0423) 8th of 8Gemini 3.5 Flash-Lite
Models writing 50–100 tokens a second 13th of 21Claude Opus 5.5 – $1.29 blended, 50th percentilecheapest: Granite 4.2 8B sets this group
Models from developers based in China 39th of 75MiMo-V2.6-Pro – $1.29 blended, 80th percentilecheapest: Qwen3.5-9B 16th of 23DeepSeek-V4.1-Flash
Open-weights models 39th of 104MiMo-V2.6-Pro – $1.29 blended, 85th percentilecheapest: Mistral Nemo 34th of 43Nemotron 3.5 Super VL

The name under each rank is the leader of that group; green is the top quarter of the group and red the bottom quarter, by rank. Price is shown as a percentile: the share of the group that costs less, so lower is cheaper. "Sets this group" marks the measure the group is defined by. Spring Prompt overall is our score out of 100 across every source (how it works). Price is the developer's list price (or the typical OpenRouter host where we haven't read one) for three input tokens to one output; intelligence and speed are from Artificial Analysis (data sourced from Artificial Analysis); "similar intelligence" means within 4 points on its Intelligence Index. Developer locations are where each company is based, not where a model is served. Groups under 3 models are not ranked.

How good is it, and for what?

Each use case ranks the available models its benchmarks measured, then those priced $1–3 per million tokens. The bar shows its percentile in that field, best to the right. Open a row for the results behind it.

Product listingsTurning a sparse product feed and photos into listings that can go live Not measured
  • CatalogBench: has not measured this model.
Decks from an analysisTurning a finished analysis into a deck you could present as it is Not measured
  • DeckBench: has not measured this model.
User surveysPlanning a user survey and reading its results without being misled Not measured
  • SurveyBench: has not measured this model.
Marketing planningPlanning a year of ad spend without overspending Not measured
  • ROASBench: has not measured this model.
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls 71st of 159 available · 16th of 31 at its price
BenchmarkQwen3.5-397B-A17BRank among availableBest available
tau2-benchtau2 v1.0.1, retail: task success (pass^1) · reported 84.4% 1st of 6 This model
tau2-benchtau2 v1.0.1, airline: task success (pass^1) · reported 81.5% 5th of 6 Claude Opus 4.5 84.0%
tau2-benchtau2 v1.0.1, telecom: task success (pass^1) · reported 97.8% 1st of 6 This model
Artificial AnalysisArtificial Analysis: Terminal-Bench 4.0 · reported 0.0% 67th of 107 Kimi K3.1 Pro 67.2%
Artificial AnalysisArtificial Analysis: Terminal-Bench 2.1 · reported 51.3% 58th of 107 Claude Fable 5.1 91.4%
Artificial AnalysisArtificial Analysis: τ-bench banking · reported 13.4% 60th of 104 Grok 4.6 50.7%
  • Berkeley Function Calling Leaderboard (BFCL) V4: pending. Its latest results are dated 16 Dec 2025, before this model was released.
  • OpenHands Index: has not measured this model.
  • Microsoft STATE-Bench: has not measured this model.
  • Vending-Bench 2: has not measured this model.
Professional workReal tasks from banking, consulting and law, business documents and freelance projects 48th of 51 available · 11th of 12 at its price
BenchmarkQwen3.5-397B-A17BRank among availableBest available
APEX-AgentsAPEX-Agents: tasks passed · reported 24.9% 36th of 37 Claude Sonnet 5.5 75.5%
  • GDP.pdf: has not measured this model.
  • Remote Labor Index: has not measured this model.
Reasoning and knowledgeHard questions across science, maths and general knowledge 82nd of 187 available · 24th of 35 at its price
BenchmarkQwen3.5-397B-A17BRank among availableBest available
Artificial AnalysisArtificial Analysis: Artificial Analysis Intelligence Index · reported 21.4 94th of 185 Claude Opus 5.5 57.6
Artificial AnalysisArtificial Analysis: Humanity's Last Exam · reported 29.0% 79th of 186 Claude Opus 5.5 61.4%
Artificial AnalysisArtificial Analysis: GPQA Diamond · reported 89.3% 50th of 168 GPT-6 Astra 96.3%
WritingWhat people prefer in blind comparisons, and judged writing quality 68th of 169 available · 18th of 28 at its price
BenchmarkQwen3.5-397B-A17BRank among availableBest available
Arena (formerly LMArena)Arena Text: overall · reported 1,442 60th of 153 Gemini 4 Argon 1,525
UGI LeaderboardUGI: writing score · reported 50.1 57th of 124 Gemini 3.8 Flash 78.6
Sticking to the factsSummarising without inventing things, and factual answers 74th of 139 available · 15th of 26 at its price
BenchmarkQwen3.5-397B-A17BRank among availableBest available
Arena (formerly LMArena)Arena Text factuality: overall · reported 1,445 58th of 109 Gemini 4 Argon 1,511
  • Vectara Hallucination Leaderboard: has not measured this model.
  • SimpleQA Verified (Epoch AI): has not measured this model.
Speed under pressureGood decisions against a real clock (fast chess) Not measured
  • BulletBench: has not measured this model.

Nearest alternatives

Against Qwen3.5-397B-A17B at $1.29 blended, Intelligence Index 21.4. "Similar intelligence" means within 4 points or better.

Where to run it

9 hosts on OpenRouter at the standard service tier, with a median of $1.29 blended per million tokens; quantisation where the host states it: FP8. Cheapest first.

HostQuantisationInputOutputBlended
Alibabanot stated$0.39$2.34$0.88
DeepInfraFP8$0.45$3.00$1.09
ParasailFP8$0.50$3.60$1.27
AtlasCloudFP8$0.55$3.50$1.29
Phalanot stated$0.55$3.50$1.29
GMICloudFP8$0.60$3.60$1.35
Novitanot stated$0.60$3.60$1.35
StreamLakenot stated$0.60$3.60$1.35

Prices per million tokens from OpenRouter's public endpoint list, checked 10 Nov 2026; the 1 dearer hosts are left out.

Against the alternatives

The models you would most likely weigh it against: the leaders of the groups above.

Scroll sideways to see every alternative.

Qwen3.5-397B-A17BMuse Spark 1.3
most intelligent at $1–3 per million tokens
DeepSeek-V4-Flash (0423)
cheapest with similar intelligence
Gemini 3.5 Flash-Lite
fastest with similar intelligence
Claude Opus 5.5
most intelligent available today
Price per million tokens $1.29 $2.00$0.15$0.85$8.00
Tokens a second 56 254–39896
Intelligence Index 21.4 48.124.422.257.6
Spring Prompt overall – 59–2582
Rank among available models, by use case
Product listings – 17th–25th13th
Decks from an analysis – 25th–29th7th
User surveys – ––19th4th
Marketing planning – 25th–28th9th
Agents and tool use 71st of 159 13th59th81st3rd
Professional work 48th of 51 8th–46th3rd
Reasoning and knowledge 82nd of 187 11th69th98th1st
Writing 68th of 169 7th63rd49th5th
Sticking to the facts 74th of 139 11th81st61st3rd
Speed under pressure – 33rd–2nd–

Shaded figures are better than Qwen3.5-397B-A17B's. A dash means no result.

Where it has been measured

27th most intelligent of the 34 models priced $1–3 per million tokens. Among models available today, in the bottom quarter for professional work.

Every result

42 published results from 5 sources, each in the source's own units, with the configuration that produced it.

APEX-Agents · reported by APEX-Agents · 1 result
MeasureValueRankConfigurationDated
APEX-Agents: tasks passed% of tasks, higher is better 24.9% 38 of 39 Qwen3.5-397B-A17B 1 Oct 2026

Not shown: Not your firm's documents or tools; graded by rubric, not by a client.

Arena (formerly LMArena) · reported by Arena (formerly LMArena) · 7 results
MeasureValueRankConfigurationDated
Arena Text factuality: overallArena rating, higher is better 1,445
1,442–1,447
81 of 182 Qwen3.5-397B-A17B 2 Oct 2026
Arena Text: business, management and financeArena rating, higher is better 1,446
1,441–1,451
85 of 406 Qwen3.5-397B-A17B 2 Oct 2026
Arena Text: creative writingArena rating, higher is better 1,405
1,400–1,411
101 of 411 Qwen3.5-397B-A17B 2 Oct 2026
Arena Text: expert promptsArena rating, higher is better 1,477
1,470–1,484
76 of 363 Qwen3.5-397B-A17B 2 Oct 2026
Arena Text: instruction followingArena rating, higher is better 1,434
1,429–1,438
90 of 413 Qwen3.5-397B-A17B 2 Oct 2026
Arena Text: overallArena rating, higher is better 1,442
1,439–1,445
92 of 413 Qwen3.5-397B-A17B 2 Oct 2026
Arena Text: writing, literature and languageArena rating, higher is better 1,421
1,416–1,426
93 of 412 Qwen3.5-397B-A17B 2 Oct 2026

Not shown: Not an error rate: claims that cannot be checked on the web are skipped, and preference still carries most of the weight.

Artificial Analysis · reported by Artificial Analysis · 23 results
MeasureValueRankConfigurationDated
Artificial Analysis: Artificial Analysis Coding Indexindex score, higher is better 48.2 103 of 187 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 18.4 225 of 413 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Artificial Analysis Intelligence Indexindex score, higher is better 21.4 195 of 413 Qwen3.5-397B-A17B (no reasoning) 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 89.3% 80 of 360 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: GPQA Diamond% of questions, higher is better 86.1% 112 of 360 Qwen3.5-397B-A17B (no reasoning) 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 29.0% 155 of 419 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Humanity's Last Exam% of questions, higher is better 19.8% 202 of 419 Qwen3.5-397B-A17B (no reasoning) 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 78.8% 10 of 288 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: IFBench% of instructions, higher is better 51.6% 134 of 288 Qwen3.5-397B-A17B (no reasoning) 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 77.3% 128 of 407 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Long-context reasoning (AA-LCR)% of questions, higher is better 64.3% 234 of 407 Qwen3.5-397B-A17B (no reasoning) 20 Oct 2026
Artificial Analysis: Output speedtokens per second, higher is better 54.7 109 of 137 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Output speedtokens per second, higher is better 55.6 106 of 137 Qwen3.5-397B-A17B (no reasoning) 20 Oct 2026
Artificial Analysis: SciCode% of problems, higher is better 44.8% 148 of 205 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Terminal-Bench 2.1% of tasks, higher is better 51.3% 106 of 188 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Terminal-Bench 4.0% of tasks, higher is better 0.0% 152 of 204 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 40.9% 51 of 282 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Terminal-Bench Hard% of tasks, higher is better 35.6% 71 of 282 Qwen3.5-397B-A17B (no reasoning) 20 Oct 2026
Artificial Analysis: Time to first answer tokenseconds, lower is better 59.9 124 of 137 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: Time to first answer tokenseconds, lower is better 1.54 36 of 137 Qwen3.5-397B-A17B (no reasoning) 20 Oct 2026
Artificial Analysis: τ-bench banking% of tasks, higher is better 13.4% 109 of 175 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 95.6% 18 of 286 Qwen3.5-397B-A17B 20 Oct 2026
Artificial Analysis: τ²-bench telecom% of tasks, higher is better 83.9% 89 of 286 Qwen3.5-397B-A17B (no reasoning) 20 Oct 2026

Not shown: Not your codebase or tools.

tau2-bench · reported by tau2-bench · 8 results
MeasureValueRankConfigurationDated
tau2 v1.0.1, airline: consistency (pass^4)% of tasks, higher is better 68.0% 4 of 8 Qwen3.5-397B-A17B (thinking reasoning, tau2) 27 Feb 2026
tau2 v1.0.1, airline: task success (pass^1)% of tasks, higher is better 81.5% 5 of 8 Qwen3.5-397B-A17B (thinking reasoning, tau2) 27 Feb 2026
tau2 v1.0.1, banking knowledge, text-emb-3-large retrieval: consistency (pass^4)% of tasks, higher is better 5.2% 1 of 2 Qwen3.5-397B-A17B (thinking reasoning, tau2) 27 Feb 2026
tau2 v1.0.1, banking knowledge, text-emb-3-large retrieval: task success (pass^1)% of tasks, higher is better 9.8% 1 of 2 Qwen3.5-397B-A17B (thinking reasoning, tau2) 27 Feb 2026
tau2 v1.0.1, retail: consistency (pass^4)% of tasks, higher is better 59.7% 1 of 8 Qwen3.5-397B-A17B (thinking reasoning, tau2) 27 Feb 2026
tau2 v1.0.1, retail: task success (pass^1)% of tasks, higher is better 84.4% 1 of 8 Qwen3.5-397B-A17B (thinking reasoning, tau2) 27 Feb 2026
tau2 v1.0.1, telecom: consistency (pass^4)% of tasks, higher is better 92.1% 1 of 8 Qwen3.5-397B-A17B (thinking reasoning, tau2) 27 Feb 2026
tau2 v1.0.1, telecom: task success (pass^1)% of tasks, higher is better 97.8% 1 of 8 Qwen3.5-397B-A17B (thinking reasoning, tau2) 27 Feb 2026

Not shown: Not your policies or customers: the tasks and policies are tau2's own, and the customer is simulated by gpt-5.2.

UGI Leaderboard · reported by UGI Leaderboard · 3 results
MeasureValueRankConfigurationDated
UGI: requested-length error% off the requested word count, lower is better 23.0% 216 of 379 Qwen3.5-397B-A17B (no reasoning) 21 Mar 2026
UGI: style adherencescore from 0 to 1, higher is better 0.33 281 of 379 Qwen3.5-397B-A17B (no reasoning) 21 Mar 2026
UGI: writing scorescore out of 100, higher is better 50.1 165 of 379 Qwen3.5-397B-A17B (no reasoning) 21 Mar 2026

Not shown: Not other format limits such as character counts or bullet counts.

Will Qwen3.5-397B-A17B work for your product?

Benchmarks aren't your task. Paste the prompt your product uses and run it beside the models you could use instead, with every answer checked and costed.

New here? Your email starts a free trial. No credit card needed.