Google Gemma 4 26B A4B Instruct
Gemma 4 26B A4B is a Mixture-of-Experts model from Google DeepMind with 26B total parameters and only 4B active per token, offering fast inference at high quality. It handles text, image, and video input, supports 256K context, function calling, and reasoning with configurable thinking modes.
Input / output
$0.13 / $0.40
per 1M tokens
Context
256K
16K max output
Reasoning
none, low, medium, high
default low
Cached input
$0.05
62% off input
Modalities
Text, image, video
out: text
Pricing
USD per 1M tokensInput
$0.13
Cached input
$0.05
Output incl. reasoning
$0.40
| Similar models | Input | Output |
|---|---|---|
| Google Gemma 4 26B A4B Instruct | $0.13 | $0.40 |
| DeepSeek V4 Flash 0423 | $0.138 | $0.275 |
| GPT-6 Luna | $0.10 | $0.50 |
| Claude Haiku 5.5 | $0.10 | $0.50 |
| GPT-4o Mini | $0.15 | $0.60 |
Prompt caching is automatic; see Prompt caching.
Specs
- Model ID
google-gemma-4-26b-a4b-it- Endpoint
POST /api/v1/chat/completions- Context window
- 256,000 tokens
- Max output
- 16,384 tokens
- Reasoning effort
reasoning_effort: none | low | medium | high· default low · guide- Served precision
- BF16
- Added
- Apr 2, 2026
- Weights
- Open
Capabilities
Modalities
- Text
- Input and output
- Image
- Input only · up to 10
- Audio
- Not supported
- Video
- Input only
Features
- Streaming (SSE)
- Supported
- Function calling
- Supported
- Structured output (json_schema)
- Supported
- Reasoning effort
- none, low, medium, high
- Prompt caching
- Supported
- Web search
- venice_parameters
- X search
- Not supported
- Log probabilities
- Supported
- Fine-tuning
- Not supported
- Image generation
- Not supported
POST
/api/v1/chat/completionsRecommendedcurl https://api.venice.ai/api/v1/chat/completions \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "google-gemma-4-26b-a4b-it",
"messages": [{ "role": "user", "content": "Explain TEE attestation in two sentences." }],
"reasoning_effort": "low"
}'Privacy
Tier 1
Anonymized
Included
Tier 2 · this model
Private
Zero data retention, enforced by contract. Prompts and outputs aren’t stored.
Tier 3
TEE
Not available
Tier 4
End-to-end encrypted
Not available