What are the hardware requirements for Codegemma 7B?

Codegemma 7B requires approximately 17.1 GB of VRAM and supports a context window of 8K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Codegemma 7B and how large is it?

Codegemma 7B is developed by google, a model with 7b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Codegemma 7B perform on standard benchmarks?

Codegemma 7B has the following published scores: MMLU 56.59, HumanEval (Python) 40.13. Compare against reference models on this page or on the LLM Explorer leaderboards.

Codegemma 7B by google — VRAM 17.1GB, 8K context

Name: Codegemma 7B
Rating: 1 (1 reviews)
Author: google

Codegemma 7B is an open-source language model by google. Features: 7b LLM, VRAM: 17.1GB, Context: 8K, License: gemma, HF Score: 56.7, LLM Explorer Score: 0.17, Arc: 53.9, HellaSwag: 76.7, MMLU: 56.6, TruthfulQA: 38, WinoGrande: 69.6, GSM8K: 45.5, HumanEval: 40.1.

Endpoints compatible Gemma Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/google/codegemma-7b

Codegemma 7B Benchmarks

ARC: 53.92 vs 96.7 (so35)^-44.2%

HellaSwag: 76.73 vs 95.3 (gpt4)^-19.5%

MMLU: 56.59 vs 88.3 (so35)^-35.9%

TruthfulQA: 38.03 vs 59 (gpt4)^-35.5%

WinoGrande: 69.61 vs 87.5 (gpt4)^-20.4%

GSM8K: 45.49 vs 96.4 (so35)^-52.8%

HumanEvalA: 30.6

Python Eva: 40.13 vs 92 (so35)^-56.4%

Java Eval: 35.03

Javascript: 43.06

CPP Eval: 40.34

PHP Eval: 34.76

Julia Eval: 28.91

D Eval: 10.44

Lua Eval: 32.88

R Eval: 20.96

Racket Eva: 13.55

Rust Eval: 34.56

Swift Eval: 32.62

LLME Score: 0.1668

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

🌟 Advertise your project 🚀

Codegemma 7B Parameters and Internals

Model Type

text-to-text, text-to-code, decoder-only

Use Cases

Areas:

Research, Commercial applications

Applications:

Code Completion, Code Generation, Code Conversation, Code Education

Primary Use Cases:

Code completion with IDE extension, Interactive code learning experiences

Limitations:

Limitations of LLMs based on training data., Potential representational harms.

Considerations:

See Gemma model card for comprehensive considerations.

Additional Notes

The model is built for Responsible AI development with a focus on open code applications.

Supported Languages

English (Fluent)

Training Details

Data Sources:

Publicly available code repositories, Open source mathematics datasets, Synthetically generated code

Data Volume:

500 billion tokens

Methodology:

FIM, PSM/SPM modes

Hardware Used:

TPUvs5e

Model Architecture:

Not explicitly mentioned

Safety Evaluation

Methodologies:

Internal red-teaming, Structured evaluations

Risk Categories:

Human safety, Representational harms, Cyber-offence capabilities

Ethical Considerations:

Testing autonomous hacking capabilities and ensuring potential harms are limited.

Responsible Ai Considerations

Fairness:

Human evaluation on prompts covering content safety and representational harms.

Transparency:

Discussions and evaluations are detailed in the Gemma model card.

Accountability:

Developed by Google, accountable for outputs under their AI principles.

Mitigation Strategies:

Controlled through structured evaluations and internal red-teaming.

Input Output

Input Format:

For pretrained model: code prefix and/or suffix for code completion and generation.

Accepted Modalities:

text

Output Format:

For instruction-tuned model: code and natural language

Performance Tips:

Ensure correct usage of FIM tokens in prompts.

LLM Name	Codegemma 7B
Repository 🤗	https://huggingface.co/google/codegemma-7b
Model Size	7b
Required VRAM	17.1 GB
Updated	2026-04-12
Maintainer	google
Model Type	gemma
Model Files	5.0 GB: 1-of-4 5.0 GB: 2-of-4 5.0 GB: 3-of-4 2.1 GB: 4-of-4
Model Architecture	GemmaForCausalLM
License	gemma
Context Length	8192
Model Max Length	8192
Transformers Version	4.39.3
Tokenizer Class	GemmaTokenizer
Padding Token	<pad>
Vocabulary Size	256000
Torch Data Type	bfloat16

Quantized Models of the Codegemma 7B

Model	Likes	Downloads	VRAM
Codegemma 7B GGUF	0	225	3 GB

Best Alternatives to Codegemma 7B

Best Alternatives	Context / RAM	Downloads	Likes
Kaggle Math Model Gemma V1	12K / 17.1 GB	5	0
Gemma 1.1 7B It	8K / 17.1 GB	20065	275
Gemma 1.1 7B It	8K / 17.1 GB	28	4
Gemma 7B It	8K / 17.1 GB	1029	10
SeaLLM 7B V2.5	8K / 17.1 GB	12410	50
Zephyr 7B Gemma DPO Avg	8K / 17.1 GB	15	0
Zephyr 7B Gemma Rpo Avg	8K / 17.1 GB	7	0
... Codegemma 2 7B It Alpaca V1.3	8K / 17.1 GB	9	1
Zephyr 7B Gemma V0.1	8K / 17.1 GB	302	124
... 7B Finetuned Sft Navarasa 2.0	8K / 34 GB	314	23

Note: green Score (e.g. "73.2") means that the model is better than google/codegemma-7b.

Rank the Codegemma 7B Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

What open-source LLMs or SLMs are you in search of? 53254 in total.

Email us: info@extractum.io. Our Privacy Policy | Terms and Conditions | Suggest an improvement.

Our Social Media →

Original data from HuggingFace, OpenCompass and various public git repos.

Check out Ag3ntum — our secure, self-hosted AI agent for server management.

Release v20260328a

Support LLM Explorer

Codegemma 7B by google