Creator

Alibaba

Released

2025-03-05

Intelligence

13.4

Artificial Analysis Index

Coding

Artificial Analysis Index

In $/1M

$0.66

input tokens

Out $/1M

$1.00

output tokens

Blended $/1M

$0.74

3:1 blended

Speed

0

tokens / sec

Profile

License, openness, and modality — the governance layer.

License

Apache 2.0

commercial OK

Weights

Open

downloadable

Modalities

Parameters

33B

total

A model focused on advancing AI reasoning capabilities, particularly excelling in mathematics and programming. Features deep introspection and self-questioning abilities while having some limitations in language mixing and recursive/endless reasoning patterns.

Capability profile

Category strength across 20 domains, via LLM Stats.

math
80structured output
80instruction following
80biology
70general
70
physics
70chemistry
70reasoning
70tool calling
70code
60
Via LLM Stats

Benchmark breakdown

Independent evaluation scores, normalized to 0–100.

GPQA Diamond
59
Humanity’s Last Exam
8
MMLU-Pro
76
SciCode
36
LiveCodeBench
63
MATH-500
96
AIME 2025
29
τ²-Bench (agentic)
Terminal-Bench Hard
IFBench
39
Via Artificial Analysis
© 2026 NYSGPT2525 LLC