Key Specifications

Vendorother
Versionjamba-1-5-large
Release Date2024-08-07
Context Window256000 tokens
Input Modalitiestext
Output Modalitiestext
LicenseJamba Open Model License
Documentationhttps://huggingface.co/models

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 84.3 % 2024-08-07 5-shot view
HUMANEVAL 73.7 pass@1 2024-08-07 view
GSM8K 83.5 % 2024-08-07 0-shot CoT view
MATH 60 % 2024-08-07 0-shot CoT view
BBH 82.8 % 2024-08-07 3-shot CoT view
GPQA 50.8 % 2024-08-07 0-shot view
IFEVAL 70.8 % 2024-08-07 prompt_strict view
ARC 93.8 % 2024-08-07 challenge view
MUSR 51.1 % 2024-08-07 0-shot view
WINOGRANDE 84.3 % 2024-08-07 0-shot view

Pricing

Tier Price Currency
Input$2 / MtokUSD
Output$8 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://huggingface.co/models · as of 2024-08-07

Compliance

  • Data Residency: self-host
  • SOC2: ✗
  • HIPAA: ✗
  • GDPR: ✗
  • ISO 27001: ✗

Jamba 1.5 Large

Model Overview

AI21 Labs Jamba 1.5 Large 混合 SSM-Transformer 模型, 256K 上下文, 总参 398B/活跃 94B, 长上下文突出。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Other jamba-1-5-large 2024-08-07 256K text text Jamba Open Model License

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 84.3 % 5-shot
HumanEval 73.7 pass@1
GSM8K (Grade School Math 8K) 83.5 % 0-shot CoT
MATH 60.0 % 0-shot CoT
BBH (BIG-Bench Hard) 82.8 % 3-shot CoT
GPQA 50.8 % 0-shot
IFEval 70.8 % prompt_strict
ARC 93.8 % challenge
MUSR 51.1 % 0-shot
WinoGrande 84.3 % 0-shot

Pricing

Input Output Cache Read Cache Write

per million tokens

Strengths

  • MMLU score 84.3, strong knowledge reasoning.
  • Context window 256K.
  • Mixture-of-Experts architecture.

Weaknesses

  • Proprietary, not self-hostable.

Use Cases

  • Code generation and debugging
  • Long document summarization

References