Inference
EN
Zyphra

ZAYA1-8B

ZAYA1-8B is a reasoning model from Zyphra in the ZAYA1 family. 9 benchmarks count toward its index, in 5 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2026-05
RankedEnough results for an index.
IndexAll categories, out of 100.38.2±8.4Median 51.0
CoverageIndex weight that has results.70%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.131KMedian 262K
EloLMArena rating and rank.—Median 1422

Run locally

8.8B parameters
16-bit
19.5 GB
8-bit
10.8 GB
4-bit
6.6 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 8 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.1 counted21.6—
CodingCode writing and repair.1 counted38.4—
ReasoningLogic problems and puzzles.None counted——
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.2 counted40.1192 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.2 counted36.846 / 63
MathMath problems.3 counted43.7116 / 168

Results

9 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
Instruction-Following EvalInstruction85.6%38 / 70
Graduate-Level Google-Proof Q&AKnowledge71.0%91 / 141
Massive Multitask Language Understanding ProfessionalKnowledge74.2%52 / 75
LiveCodeBench v6Coding65.8%45 / 59
AIME 2026Math89.1%27 / 34
GPQA Diamond · BenchLMKnowledge71.0%71 / 84
Berkeley Function Calling Leaderboard v4Agentic39.2%22 / 26
Instruction Following BenchmarkInstruction52.6%50 / 57
Harvard-MIT Mathematics Tournament February 2026Math71.6%24 / 27
IMOAnswerBenchMath59.3%9 / 10
ApexMath32.2%—

Sources

BenchLMCC BY-NC 4.0

History

5 releases
Index39.1 → 38.2
23 Oct 202612 Nov 2026

Questions

How good is ZAYA1-8B?
Its index is 38.2 out of 100. That is place 218 of 283 ranked models.
What is ZAYA1-8B best at?
Math: 43.7 out of 100, place 116 of 168.
What is the context window of ZAYA1-8B?
131K tokens.
Is ZAYA1-8B open source?
Yes. Its weights are published.
When was ZAYA1-8B released?
May 2026.

Badge

ZAYA1-8B on Inference360: place 218 of 283, index 38.2
HTML
<a href="https://inference360.ai/models/zaya1-8b"><img src="https://inference360.ai/badges/zaya1-8b.svg" alt="ZAYA1-8B on Inference360: place 218 of 283, index 38.2" height="20"></a>
Markdown
[![ZAYA1-8B on Inference360: place 218 of 283, index 38.2](https://inference360.ai/badges/zaya1-8b.svg)](https://inference360.ai/models/zaya1-8b)

Compare with

Nemotron 3 Nano Omni 30B A3B38.6

More from Zyphra

All models
ZAYA1-74B-PreviewUnrankedZAYA1-VL-8BUnrankedZonos 2Unranked