Skip to main content
This page lists all available model variants for each AI provider supported by GitWhisper. Use these exact names when specifying variants with the --model-variant flag.

Quick Reference

List variants from the command line:
# List all variants for all models
gitwhisper list-variants

# List variants for a specific model
gitwhisper list-variants --model openai
gitwhisper list-variants --model claude

OpenAI Models

Default variant: gpt-4o

GPT-5 Series (Latest)

gpt-5                    # Most advanced model
gpt-5-pro                # Enhanced version
gpt-5-mini               # Faster, cost-effective
gpt-5-nano               # Ultra-fast lightweight

GPT-4.1 Series

gpt-4.1                  # Enhanced GPT-4
gpt-4.1-mini             # Balanced speed/quality
gpt-4.1-nano             # Lightweight efficient

GPT-4o Series

gpt-4o                   # Default - optimized for speed/cost
gpt-4o-mini              # Most cost-effective
gpt-4.5-preview          # Preview of upcoming features

Realtime Series

gpt-realtime             # Low-latency responses
gpt-realtime-mini        # Fastest option

o-Series (Reasoning)

o1-preview               # Advanced reasoning
o1-mini                  # Compact reasoning
o3-mini                  # Latest reasoning model
# Use GPT-5
gitwhisper commit --model openai --model-variant gpt-5

# Use o3-mini for reasoning
gitwhisper commit --model openai --model-variant o3-mini

# Set as default
gitwhisper set-defaults --model openai --model-variant gpt-4o

Claude Models

Default variant: claude-sonnet-4-5-20250929

Claude 4 Series (Latest)

claude-sonnet-4-5-20250929        # Default - best balance
claude-sonnet-4-20250514          # Stable Sonnet 4
claude-opus-4-1-20250805          # Enhanced Opus with extended context
claude-opus-4-20250514            # Most powerful Claude

Claude 3.7 Series

claude-3-7-sonnet-20250219        # Latest 3.7 release
claude-3-7-sonnet-latest          # Auto-updates to newest 3.7

Claude 3.5 Series

claude-3-5-sonnet-latest          # Auto-updates to latest 3.5
claude-3-5-sonnet-20241022        # October 2024 release
claude-3-5-sonnet-20240620        # June 2024 release

Claude 3 Series

claude-3-opus-20240307            # Most powerful Claude 3
claude-3-sonnet-20240307          # Balanced Claude 3
claude-3-haiku-20240307           # Fast and efficient
# Use latest Sonnet 4.5
gitwhisper commit --model claude --model-variant claude-sonnet-4-5-20250929

# Use Opus 4 for maximum quality
gitwhisper commit --model claude --model-variant claude-opus-4-20250514

# Use Haiku for speed
gitwhisper commit --model claude --model-variant claude-3-haiku-20240307

Gemini Models

Default variant: gemini-2.0-flash

Gemini 2.5 Series (Latest)

gemini-2.5-pro                    # Advanced reasoning with thinking
gemini-2.5-flash                  # Updated Sep 2025, fast
gemini-2.5-flash-lite             # Most cost-efficient
gemini-2.5-flash-image            # Image generation
gemini-2.5-computer-use           # Agent interaction

Gemini 2.0 Series

gemini-2.0-flash                  # Default - optimized performance
gemini-2.0-flash-lite             # Lowest latency

Gemini 1.5 Series

gemini-1.5-pro-002                # Up to 2M tokens context
gemini-1.5-flash-002              # Up to 1M tokens context
gemini-1.5-flash-8b               # Most cost-effective
# Use default 2.0 Flash
gitwhisper commit --model gemini

# Use 2.5 Pro with thinking mode
gitwhisper commit --model gemini --model-variant gemini-2.5-pro

# Use 1.5 Pro for massive context
gitwhisper commit --model gemini --model-variant gemini-1.5-pro-002

Grok Models

Default variant: grok-2-latest

Grok 4 Series

grok-4                            # Most intelligent
grok-4-heavy                      # Most powerful
grok-4-fast                       # Efficient reasoning

Grok Code Series

grok-code-fast-1                  # Agentic coding

Grok 3 & 2 Series

grok-3                            # Reasoning capabilities
grok-3-mini                       # Faster responses
grok-2-latest                     # Default - stable release
# Use Grok 4
gitwhisper commit --model grok --model-variant grok-4

# Use Grok for coding
gitwhisper commit --model grok --model-variant grok-code-fast-1

Llama Models

Default variant: llama-3-70b-instruct

Llama 3.x Series

llama-3.3-70b-instruct            # Latest 70B model
llama-3.2-3b-instruct             # Small efficient model
llama-3.2-1b-instruct             # Tiny efficient model
llama-3.1-405b-instruct           # Largest Llama model
llama-3.1-70b-instruct            # Large balanced model
llama-3.1-8b-instruct             # Small fast model

Llama 3.0 Series

llama-3-70b-instruct              # Default - proven quality
llama-3-8b-instruct               # Fast and efficient
# Use Llama 3.3 70B
gitwhisper commit --model llama --model-variant llama-3.3-70b-instruct

# Use small model for speed
gitwhisper commit --model llama --model-variant llama-3.1-8b-instruct

DeepSeek Models

Default variant: deepseek-chat

DeepSeek V3 Series

deepseek-v3.2-exp                 # Latest experimental
deepseek-v3.1                     # Hybrid reasoning
deepseek-v3.1-terminus            # Terminus variant
deepseek-v3-0324                  # March 2024 release

DeepSeek R1 & Chat

deepseek-r1-0528                  # Upgraded reasoning
deepseek-reasoner                 # Reasoning-focused
deepseek-chat                     # Default - general purpose
# Use DeepSeek Chat
gitwhisper commit --model deepseek

# Use reasoning model
gitwhisper commit --model deepseek --model-variant deepseek-r1-0528

GitHub Models

Default variant: gpt-4o

Available Models

gpt-4o                            # Default - OpenAI's flagship
DeepSeek-R1                       # DeepSeek reasoning
Llama-3.3-70B-Instruct            # Meta Llama
Deepseek-V3                       # DeepSeek V3
Phi-4-mini-instruct               # Microsoft Phi
"Codestral 25.01"                 # Mistral code model (quote the name)
"Mistral Large 24.11"             # Mistral large (quote the name)
Some model names contain spaces and must be quoted when used on the command line.
# Use GPT-4o (default)
gitwhisper commit --model github

# Use DeepSeek for reasoning
gitwhisper commit --model github --model-variant DeepSeek-R1

# Use Codestral (note the quotes)
gitwhisper commit --model github --model-variant "Codestral 25.01"

Ollama Models

Default variant: llama3.2:latest

Ollama Variants

Ollama supports hundreds of models. You can use any model from ollama.com/library. Popular choices:
llama3.2:latest                   # Default - balanced choice
codellama                         # Code-optimized
deepseek-coder                    # Excellent for code
qwen2.5-coder                     # Qwen code model
phi3                              # Fast and efficient
gemma2                            # Google's model
mixtral                           # High quality
llama3:70b                        # Large powerful model
# First, pull the model locally
ollama pull codellama

# Then use with GitWhisper
gitwhisper commit --model ollama --model-variant codellama

# Use specific version
gitwhisper commit --model ollama --model-variant llama3.2:3b
See the Ollama documentation for more details on installing and using local models.

Free Model

Default variant: default (only option)

No Variants

The free model doesn’t support variant selection - it uses a single default model from LLM7.io:
# Simply use the free model
gitwhisper commit --model free

# No variants available
# --model-variant flag is ignored
The free model is powered by LLM7.io and requires no API key. See Free Model documentation for limits and details.

Setting Default Variants

You can set a default model and variant to avoid specifying them each time:
# Set default model only (uses default variant)
gitwhisper set-defaults --model openai

# Set default model with specific variant
gitwhisper set-defaults --model claude --model-variant claude-opus-4-20250514

# Set for Ollama with custom base URL
gitwhisper set-defaults --model ollama --model-variant codellama --base-url http://localhost:11434

# View current defaults
gitwhisper show-defaults

# Clear defaults
gitwhisper clear-defaults

Comparison Table

Default Variants Summary

ProviderDefault VariantQualitySpeedCost
OpenAIgpt-4o⭐⭐⭐⭐⭐⭐⭐⭐⭐💰💰
Claudeclaude-sonnet-4-5-20250929⭐⭐⭐⭐⭐⭐⭐⭐⭐💰💰
Geminigemini-2.0-flash⭐⭐⭐⭐⭐⭐⭐⭐⭐💰
Grokgrok-2-latest⭐⭐⭐⭐⭐⭐⭐💰💰
Llamallama-3-70b-instruct⭐⭐⭐⭐⭐⭐💰💰
DeepSeekdeepseek-chat⭐⭐⭐⭐⭐⭐⭐💰
GitHubgpt-4o⭐⭐⭐⭐⭐⭐⭐⭐Free (limited)
Ollamallama3.2:latest⭐⭐⭐⭐⭐⭐Free (local)
Freedefault⭐⭐⭐⭐⭐⭐Free (limited)

Tips for Choosing Variants

Recommended defaults:
  • OpenAI: gpt-4o or gpt-4o-mini
  • Claude: claude-sonnet-4-5-20250929
  • Gemini: gemini-2.0-flash
These provide excellent quality at reasonable cost.
Best quality variants:
  • OpenAI: gpt-5 or o3-mini
  • Claude: claude-opus-4-20250514
  • Gemini: gemini-2.5-pro
Use for important commits or complex changes.
Most economical:
  • OpenAI: gpt-4o-mini
  • Claude: claude-3-haiku-20240307
  • Gemini: gemini-2.0-flash-lite
  • Free: free model
  • Local: Ollama models
Great for high-volume usage or testing.
Keep code local:
  • Ollama: Any model (codellama, llama3.2, etc.)
Your code never leaves your machine.

Staying Updated

Model variants are regularly updated:
# Check current variants
gitwhisper list-variants

# Check for specific model
gitwhisper list-variants --model claude

# Update GitWhisper to get latest models
gitwhisper update
New models are added regularly. Check the GitHub repository for announcements.

Next Steps

Model Overview

Learn how to choose the right model

OpenAI Guide

Deep dive into OpenAI models

Claude Guide

Explore Claude model options

Configuration

Set up your default preferences

Build docs developers (and LLMs) love