Qwen3.5 35B A3B — Benchmarks

Benchmark scores for Qwen3.5 35B A3B aggregated from public leaderboards (9 boards), with how it ranks among open models and how it compares with its nearest neighbours. See hardware requirements for what you need to run it.

llmrun Composite Score →

105.8

90% range 101.6–110.1 · from 7 boards

CodingNot enough data yet
AgentsNot enough data yet
MathNot enough data yet
Science107.690% range 99–116
Reasoning106.790% range 98–116

Qwen3.5 35B A3B reasoning benchmarks

Multi-step logic and problem-solving — can the model think through a hard problem rather than recall an answer.

GPQA Diamond
GPQA Diamond is a set of extremely hard, graduate-level science questions (physics, chemistry, biology) written by domain experts and filtered so that skilled non-experts with web access still fail. It measures genuine reasoning rather than memorization.
83.5%
83.5%
median 55.7%open leader 93.2%
Overall best: Gemini 4 Argon (high) 96.4%
76th percentile · #25 of 100 open · #108 of 309 all
#25 of 100
#108 of 309
Independent
Epoch AI · Feb 2026
CritPt
CritPt (Critical Points) tests graduate-level physics problem-solving on research-style questions spanning mechanics, electromagnetism, quantum mechanics, and more. It was created by physicists to probe whether a model can reason through multi-step physics rather than pattern-match textbook answers.
0.6%
0.6%
median 0.9%open leader 26.6%
Overall best: Gemini 4 Carbon (high) 36.0%
47th percentile · #45 of 84 open · #180 of 244 all
#45 of 84
#180 of 244
Lab-reported
Epoch AI · Feb 2026
DTBench
DTBench is a set of handcrafted, expert-validated multiple-choice questions on decision theory — Newcomb-like problems and related puzzles — built by the Conceptual Reasoning Index project with Anthropic. It tests whether a model can follow an abstract argument rather than recall a fact.
80.0%
80.0%
median 68.7%open leader 93.9%
Overall best: Claude Opus 5.5 Max 98.9%
67th percentile · #24 of 70 open · #138 of 235 all
#24 of 70
#138 of 235
Lab-reported
Epoch AI · Feb 2026
LMCA
LMCA (Language Model Conceptual Argumentation) scores models on expert-rated arguments across philosophy, decision theory, and AI risk — domains where there is no empirical feedback loop to check an answer against. It is part of the Conceptual Reasoning Index, built with Anthropic by researchers Emery Cooper and Caspar Oesterheld.
29.5%
29.5%
median 25.5%open leader 55.5%
Overall best: Claude Sonnet 5.5 Max 68.3%
60th percentile · #23 of 56 open · #140 of 197 all
#23 of 56
#140 of 197
Lab-reported
Epoch AI · Feb 2026

Qwen3.5 35B A3B coding benchmarks

Writing, editing, and fixing real code — measured by whether the resulting program actually runs and passes tests.

SciCode
SciCode asks a model to write real scientific-computing code — implementing multi-step numerical methods drawn from physics, chemistry, biology, and math research — and checks the output against reference solutions. It was built by a team of scientists and ML researchers to test coding ability on genuine research problems, not typical software tasks.
29.3%
29.3%
median 40.1%open leader 60.9%
Overall best: Claude Opus 5.5 Max 66.9%
17th percentile · #60 of 72 open · #201 of 217 all
#60 of 72
#201 of 217
Lab-reported
Epoch AI · Feb 2026

Other results (1)leaderboard-only, not part of the llmrun Score

Leaderboard-only, not part of the llmrun Score

LMArena WebDev
Code Arena (formerly WebDev Arena) is Arena's head-to-head coding leaderboard: two models each build a real web app or UI from the same prompt, and people vote blind on which result is better. The rating shown is a relative Bradley-Terry score, not a percentage — compare it only to other models on this board.
1254.2
1254.2
median 1415.2open leader 1665.0
Overall best: Claude Opus 5.5 (Max) 1809.8
7th percentile · #51 of 55 open · #133 of 146 all
#51 of 55
#133 of 146
Arena
Arena · Oct 2026

Qwen3.5 35B A3B math benchmarks

Symbolic and quantitative problem-solving, from competition math to multi-step calculation.

AIME 2024/2025
AIME (American Invitational Mathematics Examination) is a prestigious high-school competition of hard, integer-answer problems. It's a widely-cited yardstick for multi-step mathematical reasoning, here on the 2024–2025 papers.
70.0%
70.0%
median 54.4%open leader 99.4%
Overall best: Claude Fable 5 (high) 100.0%
66th percentile · #31 of 88 open · #151 of 292 all
#31 of 88
#151 of 292
Independent
Epoch AI · Feb 2026

Qwen3.5 35B A3B knowledge benchmarks

Breadth and depth of factual knowledge across many subjects, paired with the reasoning to apply it.

MMLU-Pro
MMLU-Pro is a harder, cleaned-up successor to MMLU with ten answer choices and more reasoning-heavy questions across 14 subjects, measuring broad knowledge and reasoning together.
85.3%
85.3%
median 52.0%open leader 88.0%
Overall best: Gemini-3.1-Pro 91.2%
96th percentile · #7 of 146 open · #32 of 257 all
#7 of 146
#32 of 257
Independent
TIGER-Lab

Qwen3.5 35B A3B instruction following benchmarks

How precisely a model obeys explicit, verifiable instructions — format, length, and content constraints — and, more broadly, how much people prefer its responses head-to-head in blind human voting.

LMArena Text
Arena (arena.ai, formerly LMArena / Chatbot Arena) ranks models by millions of blind, head-to-head human votes on real prompts. The Arena Score shown here is a Bradley-Terry rating computed with Arena's default style control, so higher means people preferred that model's answers more often — it's a relative rating, not a percentage, so compare it only to other models on this board.
1393.8
1393.8
median 1321.4open leader 1488.2
Overall best: Gemini 4 Carbon (High) 1527.5
71st percentile · #55 of 188 open · #166 of 408 all
#55 of 188
#166 of 408
Arena
Arena · Oct 2026

Score vs VRAM in the 24 GB tier

Qwen3.5 35B A3B
Other models
Best Score for its VRAM
x: VRAM (GB) · y: llmrun Score

#6 of 36 that fit 24 GB

Each dot is a model at the best quant that fits a 24 GB GPU. The line steps up to the best Score available at each amount of VRAM; a dot below it is beaten by a model that needs no more memory.

Full 24 GB ranking →

Qwen3.5 35B A3B vs its neighbours

Same boards for the nearest open models per llmrun Score and the best model that fits 24 GB. The highest score in each row is marked.

Board	Qwen3.5 35B A3Bthis model	Qwen3.8 27Bbest in classQ4_K_M · 18 GB	Qwen3.6 35B A3Bnearest by ScoreQ4_K_M · 22 GB	Gemma 4 31B ITnearest by ScoreQ4_K_M · 22 GB	Gemma 4 26B A4B ITnearest by ScoreQ4_K_M · 17 GB
llmrun Score	105.8	122.0	106.8	106.7	105.8
GPQA Diamond	83.5%	—	84.9%	75.8%	73.2%
CritPt	0.6%	5.4%	0.3%	1.4%	0.0%
DTBench	80.0%	88.0%	73.9%	82.7%	74.9%
LMCA	29.5%	41.4%	29.7%	39.3%	29.7%
SciCode	29.3%	46.6%	36.6%	45.5%	40.1%
AIME 2024/2025	70.0%	—	86.7%	73.3%	82.2%
LMArena Text	1393.8	1435.6	—	1452.3	1437.6
Compare head to head		Compare	Compare	Compare	Compare

Scores aggregated from public benchmark sources (Epoch AI: 6 boards, Arena: 2 boards, TIGER-Lab: 1 board) — each linked from the benchmark pages. llmrun does not run these benchmarks. Percentiles and ranks are among open models in the llmrun catalogue; the median and leader marks refer to the same field.