Benchmarks of Apriel v1.5 15B Thinker14
BenchmarkByScoreRank
Overall1
Knowledge4
Science2
Coding1
Multimodal1
Documents1
Legacy · saturated or no longer reported4
No benchmark matches
—per task?
self-host onlyprice
—output speed?
128Kcontext
Ranking?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-6 Lunalow+67%—$0.005Function calling only with none · Chat Completions+3
OpenQwen3.8 27Bxhigh×2.8—$1.01Memory (4-bit): 7.4 GB → 17 GB+5BetterGPT-5.6 Lunalow+63%—$0.01
BetterMiMo v2.5thinkingOpen×2.2—$0.019Memory (4-bit): 7.4 GB → 186 GB
BetterClaude Haiku 5.5low×2.2—$0.024Remove sampling parameters+5ModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-6 Lunalow+67%—$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalow+63%—$0.01MiMo v2.5thinkingOpen×2.2—$0.019Memory (4-bit): 7.4 GB → 186 GBClaude Haiku 5.5low×2.2—$0.024Remove sampling parameters+5
Mercury 2high+41%—$0.039No image inputModelScoreCostPer taskWhat changes
Qwen3.8 27BxhighOpen×2.8—$1.01Memory (4-bit): 7.4 GB → 17 GB+5Qwen 3.5 27BthinkingOpen+97%—$0.30 / $2.40Memory (4-bit): 7.4 GB → 17 GB
Gemma 4 31BthinkingOpen+95%—$0.10 / $0.34Memory (4-bit): 7.4 GB → 19 GBQwen3.6 27BthinkingOpen+90%—$0.62Memory (4-bit): 7.4 GB → 17 GBQwen3.6 35B A3BthinkingOpen+83%—$0.48Memory (4-bit): 7.4 GB → 22 GBRun it yourself
estimateFits on✓ 8 GB (phone, laptop) · compressed✓ 16 GB · 8-bit✓ 24 GB GPU · 8-bit✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceMITfree for commercial use
History
2 versionsBenchmarks
all categoriesChat & Knowledge
Humanity's Last Exam12.1#160
AA Omniscience Index-54.8#237
CritPt (physics)1.1#120
AA Intelligence Index 4.3.2≈13.8#156
AA Omniscience · Accuracy16.5#209
AA-LCR (long context) 1.121.3#236
GPQA Diamond71.3#180
AIME 2025 · AA archived87.5#27
Coding
LiveCodeBench · AA archived72.8#37
Agents & Tool Use
τ²-bench · AA68.4#96
Writing & Language
IFBench (instructions)61.7#81
Vision & Documents
MMMU-Pro (vision)57.1#102
FAQ
How much does Apriel v1.5 15B Thinker cost?›
There is no per-unit API price: self-host only.
Is Apriel v1.5 15B Thinker free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Apriel v1.5 15B Thinker, and what is it best at?›
#159 of 334 in Chat & Knowledge, #121 of 239 in Coding, #94 of 242 in Agents & Tool Use, #81 of 245 in Writing & Language, #100 of 124 in Vision & Documents.
Is Apriel v1.5 15B Thinker better than Gemini 2.5 Flash?›
On Humanity's Last Exam, Apriel v1.5 15B Thinker scores 12.1 vs 12.1 — about the same.
What is the context window of Apriel v1.5 15B Thinker?›
128K tokens.
When was Apriel v1.5 15B Thinker released? Is there a newer version?›
Released 2025-09-30. It has been superseded by Apriel v1.6 15B Thinker.
Is Apriel v1.5 15B Thinker open source?›
Yes, the weights are open (MIT licence).
Can I run Apriel v1.5 15B Thinker locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Apriel v1.5 15B Thinker read images?›
It accepts text and images as input.
What is Apriel v1.5 15B Thinker?›
Apriel v1.5 15B Thinker is an open-weights language model from ServiceNow, released Sep 2025.
What are the best alternatives to Apriel v1.5 15B Thinker?›
GPT-6 Luna (better), Qwen3.8 27B (open weights).