Compare / Alibaba vs Meta

AI labs compared · October 2026

Alibaba vs Meta: which AI lab is ahead?

Each lab's best model on sale goes head to head on every kind of work we measure, from our own benchmarks and licensed sources. 36 models in all; a lab is only as good as the model you would actually pick from it.

27 models on sale with our resultsAlibaba Best: Qwen3.8-Max (0902) · newest: Qwen3.8-27B

9 models on sale with our resultsMeta Best: Muse Spark 1.3 · newest: Muse Spark 1.3

Alibaba wins 3Meta wins 5

Kinds of work led by each lab's best model there, of the 8 both are measured on.

Results as of 8 October 2026 (catalogue release 2026-10-08-4199ce0620fe). Models on sale means listed on OpenRouter today.

Alibaba or Meta? The short answer

Meta is ahead: its best models beat Alibaba's in 5 of the 8 kinds of work both are measured on; Alibaba leads in 3.

Pick Alibaba for decks from an analysis, agents and tool use and sticking to the facts.
Pick Meta for product listings, marketing planning, professional work, reasoning and knowledge and writing.
On our overall leaderboard, Muse Spark 1.3 is #6 (65 of 100) and Qwen3.8-Max (0902) is #12 (48).
Prices start at $0.11 per million tokens for Alibaba (Qwen3.5-9B) and $0.058 for Meta (Llama 3.1 8B Instruct).

Head to head

48Qwen3.8-Max (0902)

Best Spring Prompt overall score

65Muse Spark 1.3

45.4Qwen3.8-Max (0902)

Best Intelligence Index

48.1Muse Spark 1.3

$3.00Qwen3.8-Max (0902)

Flagship price, per million tokens

$2.00Muse Spark 1.3

$0.11Qwen3.5-9B

Cheapest model, per million tokens

$0.058Llama 3.1 8B Instruct

Models on sale with our results

Each line names the model behind the number. Prices are blended, three input tokens to one output, at the developer's list price where we have read it; see every model's price. Intelligence Index from Artificial Analysis; overall score from our leaderboard.

Which lab is better for what

For each kind of work, each lab's best model there and its rank among every model on sale. Further right is better; the better rank is in green.

Product listingsTurning a sparse product feed and photos into listings that can go live

17 of 20Qwen3.8-Max (0902) 9 of 20Muse Spark 1.3

Decks from an analysisTurning a finished analysis into a deck you could present as it is

11 of 19Qwen3.8-Max (0902) 14 of 19Muse Spark 1.3

Marketing planningPlanning a year of ad spend without overspending

14 of 19Qwen3.8-Max (0902) 12 of 19Muse Spark 1.3

Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls

7 of 164Qwen3.8-Max (0902) 9 of 164Muse Spark 1.3

Professional workReal tasks from banking, consulting and law, business documents and freelance projects

18 of 51Qwen3.8-Max (0902) 8 of 51Muse Spark 1.3

Reasoning and knowledgeHard questions across science, maths and general knowledge

17 of 198Qwen3.8-Max (0902) 8 of 198Muse Spark 1.3

WritingWhat people prefer in blind comparisons, and judged writing quality

16 of 188Qwen3.8-Max (0902) 5 of 188Muse Spark 1.3

Sticking to the factsSummarising without inventing things, and factual answers

6 of 152Qwen3-8B 11 of 152Muse Spark 1.3

Speed under pressureGood decisions against a real clock (fast chess)

not measured 19 of 22Muse Spark 1.3

Every model, by intelligence and price

Each dot is a model on sale with an Intelligence Index from Artificial Analysis. Up is smarter; left is cheaper. Hover or tap a dot for its name.

AlibabaMeta

On our overall leaderboard

The 2 of their models we rank overall: one score out of 100 across everything we publish (how the score works). Only models with enough of our own results are ranked.

#6Muse Spark 1.365
#12Qwen3.8-Max (0902)48

Benchmark by benchmark

The headline result of every benchmark both labs have models on: each lab's best model and its rank among the models on sale. Green marks the better rank.

Measure	Alibaba	Meta
Agents and tool use
Berkeley Function Calling Leaderboard (BFCL) V4 · Overall accuracy · BFCL% correct, higher is better	52.2%Qwen3-235B-A22B-Instruct-2507 · 13 of 38	31.9%Llama 3.3 70B Instruct · 26 of 38
Artificial Analysis · Terminal-Bench 4.0% of tasks, higher is better	38.9%Qwen3.8-Max (0902) · 11 of 105	33.3%Muse Spark 1.3 · 14 of 105
Artificial Analysis · Terminal-Bench 2.1% of tasks, higher is better	88.8%Qwen3.8-Max (0902) · 5 of 115	85.4%Muse Spark 1.3 · 10 of 115
Artificial Analysis · τ-bench banking% of tasks, higher is better	48.0%Qwen3.8-27B · 5 of 113	50.5%Muse Spark 1.3 · 2 of 113
Vending-Bench 2 · Money after a yearUS dollars, higher is better	$5,115Qwen3.6-Plus · 24 of 56	$6,520Muse Spark 1.1 · 13 of 56
Decks from an analysis
DeckBench · Deck ratingrating, higher is better	1,100Qwen3.8-Max (0902) · 11 of 19	821Muse Spark 1.3 · 14 of 19
Marketing planning
ROASBench · Overall scorescore out of 100, higher is better	25.4Qwen3.8-Max (0902) · 14 of 19	34.1Muse Spark 1.3 · 12 of 19
Product listings
CatalogBench · Reliably publish-ready% of products, higher is better	28.6%Qwen3.8-Max (0902) · 13 of 20	48.2%Muse Spark 1.3 · 7 of 20
CatalogBench · Reliably publish-ready · sales brief% of products, higher is better	0.0%Qwen3.8-Max (0902) · 18 of 20	10.7%Muse Spark 1.3 · 9 of 20
Professional work
APEX-Agents · Tasks passed% of tasks, higher is better	24.9%Qwen3.5-397B-A17B · 36 of 37	57.8%Muse Spark 1.3 · 12 of 37
GDP.pdf · Professional document tasks% of rubric, higher is better	23.2%Qwen3.8-Max (0902) · 13 of 31	27.6%Muse Spark 1.3 · 5 of 31
Reasoning and knowledge
Artificial Analysis · Artificial Analysis Intelligence Indexindex score, higher is better	45.4Qwen3.8-Max (0902) · 13 of 198	48.1Muse Spark 1.3 · 8 of 198
Artificial Analysis · Humanity's Last Exam% of questions, higher is better	43.1%Qwen3.8-Max (0902) · 23 of 197	48.7%Muse Spark 1.3 · 10 of 197
Artificial Analysis · GPQA Diamond% of questions, higher is better	92.8%Qwen3.8-Max (0902) · 15 of 187	94.1%Muse Spark 1.3 · 5 of 187
Sticking to the facts
Vectara Hallucination Leaderboard · Hallucination rate · Vectara% of summaries, lower is better	4.8%Qwen3-8B · 5 of 79	7.7%Llama 4 Scout · 27 of 79
Arena (formerly LMArena) · Overall · Text factualityArena rating, higher is better	1,472Qwen3.8-Max (0902) · 18 of 113	1,483Muse Spark 1.2 · 9 of 113
SimpleQA Verified (Epoch AI) · Correct answers% of questions, higher is better	55.8%Qwen3.7-Max · 18 of 73	60.3%Muse Spark 1.2 · 15 of 73
Writing
Arena (formerly LMArena) · Overall · TextArena rating, higher is better	1,482Qwen3.8-Max (0902) · 17 of 166	1,494Muse Spark 1.3 · 7 of 166
UGI Leaderboard · Writing score · UGIscore out of 100, higher is better	58.2Qwen3.6-Plus · 38 of 138	26.7Llama 4 Scout · 120 of 138

Their models, one against one

Qwen3.8-Max (0902) vs Muse Spark 1.3Benchmarks, price and which to pick

Quick answers

Is Alibaba or Meta better?

Meta is ahead: its best models beat Alibaba's in 5 of the 8 kinds of work both are measured on; Alibaba leads in 3.

What is Alibaba's best model?

Qwen3.8-Max (0902): #12 on the Spring Prompt overall leaderboard and 45.4 on the Artificial Analysis Intelligence Index.

What is Meta's best model?

Muse Spark 1.3: #6 on the Spring Prompt overall leaderboard and 48.1 on the Artificial Analysis Intelligence Index.

Which is better for agents and tool use, Alibaba or Meta?

Alibaba: its best model there, Qwen3.8-Max (0902), ranks 7 of 164 models on sale, against 9 for Meta's Muse Spark 1.3.

Which is better for reasoning and knowledge, Alibaba or Meta?

Meta: its best model there, Muse Spark 1.3, ranks 8 of 198 models on sale, against 17 for Alibaba's Qwen3.8-Max (0902).

Which is better for writing, Alibaba or Meta?

Meta: its best model there, Muse Spark 1.3, ranks 5 of 188 models on sale, against 16 for Alibaba's Qwen3.8-Max (0902).

Which is cheaper, Alibaba or Meta?

Meta's cheapest model with results, Llama 3.1 8B Instruct, costs $0.058 per million tokens (three input to one output), against $0.11 for Qwen3.5-9B from Alibaba.

Other labs compared

Which model is best for your product?

Benchmarks aren't your data. We test the models from Alibaba, Meta and the rest on your own tasks, with the cost of every answer.

Talk to us See the leaderboard