📡 LLM Pulse|🆕 New on leaderboard: Baidu: ERNIE-Light-DT-1.0 by Baidu — $1.13/M tokensLearn more →1/10

Measurement sources and verification status (13 fields)

Collected third-party figures are not independent tests. A source link and collection date do not establish reproducibility. Exact model version, dataset and test settings may be missing. No independent verification date is supplied.

speed: 160

latency: 0.28

contextWindow: 81920

inputPrice: 0.2

outputPrice: 2.4

benchmarkMmlu: 86

benchmarkHumanEval: 86

benchmarkMath: 90

benchmarkHellaSwag: 87

benchmarkGpqa: 64

benchmarkMmluPro: 62

reasoningScore: 77

codingScore: 82

Qwen: Qwen3 30B A3B Thinking 2507

qwen🔒 Proprietary

Qwen3-30B-A3B-Thinking-2507 is a 30B parameter Mixture-of-Experts reasoning model optimized for complex tasks requiring extended multi-step thinking. The model is designed specifically for “thinking mode,” where internal reasoning traces are separated...

LLM Stats Scores

77.0

Reasoning

82.0

Coding

🎯

64.0%GPQA Diamond — graduate-level science accuracy

62.0%MMLU-Pro

Published benchmark evidence

Task scores are reported independently by each publisher; they are not combined into a hidden quality score.

View HLE source ↗

10.3%

Humanity's Last Exam

Quality Score

80.0

Overall (0–100)

Speed

160 c/s

characters/second

Latency (TTFT)

0.28s

to first token

Input Price

$0.2

per 1M tokens

Output Price

$2.4

per 1M tokens

Context Window

82K

tokens

Release Date

2025-08-28

30-Day Rank History

📊 Benchmark Scores

Reported scores · % accuracy

What do these benchmarks test?

🎯

GPQA Diamond64.0%Below avg

Designed so that even PhD-level human experts average ~65–70%. A score above 80% signals near-expert scientific reasoning that goes far beyond memorisation.

📚

MMLU-Pro62.0%Below avg

The harder successor to MMLU — more options per question eliminate single-answer guessing. Tests depth of professional knowledge across 14 disciplines.

🧠

MMLU86.0%Solid

The classic academic breadth benchmark. Top frontier models now score above 85%, so it best distinguishes mid-tier from strong generalist models.

💻

HumanEval86.0%Solid

Measures whether a model writes functionally correct code — not just plausible-looking syntax. Strong scores (>85%) indicate reliable day-to-day coding assistance.

∑

MATH Benchmark90.0%Top tier

Tests multi-step mathematical reasoning under strict correctness. Even a 60% score requires strong chain-of-thought; top reasoning models now exceed 90%.

🌐

HellaSwag87.0%Solid

Tests grounded, everyday common sense rather than memorised facts. Most frontier models now score 90%+, so it best separates small/mid models from large ones.