Apriel-v1.6-15B-Thinker

ServiceNow·Released 2025-11-25

Compare

Related Models

Apriel-v1.5-15B-Thinker2025-09-30

Quality Index

27.6

168th of 507

Top 33%

Coding Index

22.0

183rd of 417

Top 44%

Math Index

88.0

37th of 269

Top 14%

Speed

69 tok/s

Top 61%

TTFT

0.20s

Market Position

Apriel-v1.6-15B-ThinkerMarket Average

Performance

69

tokens/sec

Faster than 39% of models

0.20

seconds

Faster than 99% of models

29.24

seconds

Faster than 16% of models

Market Median

86 tok/s

20% slower

Median TTFT

1.07s

81% faster

Speed Comparison

Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

69 tok/s-0%

GPT-5 (medium)

69 tok/s-0%

Qwen3 235B A22B (Reasoning)

69 tok/s-1%

Benchmarks

MMLU-Pro

79.0%

GPQA Diamond

73.3%

HLE

9.8%

LiveCodeBench

80.7%

SciCode

37.3%

TerminalBench Hard

14.4%

MATH-500Not evaluated

AIMENot evaluated

AIME 2025

88.0%

IFBench

69.1%

Long Context Recall

50.3%

Tau2

69.3%

Market AverageTop Score

Quick Compare

Similar Models

Qwen3 VL 235B A22B (Reasoning)

Alibaba

Q: 27.6$2.17/1M

Slower: 49%

DeepSeek V3.1 (Reasoning)

DeepSeek

Q: 27.7$0.86/1M

Coding: +7.7

Mistral Small 4 (Reasoning)

Mistral

Q: 27.8$0.26/1M

Faster: 139%

GPT-5.1 (Non-reasoning)

OpenAI

Q: 27.4$3.44/1M

Faster: 71%Coding: +5.3

Qwen3.5 9B (Non-reasoning)