gemma-3-4b-it
deepinfra · chat model
gemma-3-4b-it is listed here as a chat model from deepinfra. This page shows simple API pricing, token limits, and capability flags so you can compare it with similar options.
Input
$0.0400 / 1M tokens
Output
$0.0800 / 1M tokens
Cached input
N/A
Context
131.1K
Quick read
Best for
Use this page when you need a fast view of cost, context size, and supported features before testing the model in your own workload.
Things to verify
Always check the provider page for discounts, cache pricing, region rules, and any model limits that may not appear in public metadata.
Pricing
| Item | Price |
|---|---|
| Input | $0.0400 / 1M tokens |
| Output | $0.0800 / 1M tokens |
| Embedding | $0.0400 / 1M tokens |
Token limits
Context window
131.1K
Max input tokens
131.1K
Max output tokens
131.1K
Max tokens
131.1K
Capabilities
| Capability | Supported |
|---|---|
| Vision | — |
| Function calling | ✅ |
| Parallel function calling | — |
| Tool choice | ✅ |
| Prompt caching | — |
| Reasoning | — |
| Response schema | — |
| System messages | — |
| Audio input | — |
| Audio output | — |
| Web search | — |
| PDF input | — |
| Video input | — |
Benchmarks
Most benchmark rows are attached to the base model family rather than this provider route. Open benchmark explorer
| Benchmark | Score | Metric | Scope | Checked | Source |
|---|---|---|---|---|---|
| MMLU | 78.6 | 5-shot | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| MMLU-Pro (CoT) | 52.2 | 5-shot | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| GPQA | 24.3 | 5-shot | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| HumanEval | 48.8 | 0-shot | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| MATH | 50.0 | 4-shot | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| MMLU | 78.6 | 5-shot | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| GSM8K | 82.6 | 8-shot | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| HumanEval | 48.8 | 0-shot | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| MGSM | 74.3 | 0-shot | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| DocVQA | 85.6 | val | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| MMMU | 56.1 | pt | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| DocVQA | 85.6 | score | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| InfoVQA | 59.4 | score | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| MMMU | 56.1 | score | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| TextVQA | 68.6 | score | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| AI2D | 79.0 | score | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
| ChartQA | 76.3 | score | Base model: Gemma 3 (Gemma 3 PT 27B) | 2026-05-31 | Link |
Sources
| Source links | |
| Pricing data | LiteLLM model cost map |
| Synced at | 2026-05-28 |
Docs
| Official docs |
Similar models
This list is ranked by overall similarity. Use filters to emphasize the lens that matters most for the replacement you are making.
Comparing from
| Model | Cost | Input shape | Features | Context | Why it is close |
|---|---|---|---|---|---|
| gemma-3-4b-it deepinfra | In $0.0400 / 1M tokens Out $0.0800 / 1M tokens | imagetext
Output: text | Function callingTool choice | 131.1K | Current model Reference row |
| Model | Cost | Input shape | Features | Context | Why it is close |
|---|---|---|---|---|---|
| gemma-3-12b-it deepinfra | In $0.0500 / 1M tokens Out $0.1000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 131.1K | Same provider Overall 96% |
| gemma-3-27b-it deepinfra | In $0.0900 / 1M tokens Out $0.1600 / 1M tokens | imagetext
Output: text | Function callingTool choice | 131.1K | Same provider Overall 90% |
| Mistral-Small-3.2-24B-Instruct-2506 deepinfra | In $0.0750 / 1M tokens Out $0.2000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 128.0K | Same provider Overall 89% |
| ministral-3b-2512 OpenRouter | In $0.1000 / 1M tokens Out $0.1000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 131.1K | Text + image covered Overall 84% |
| gpt-oss-20b deepinfra | In $0.0400 / 1M tokens Out $0.1500 / 1M tokens | imagetext
Output: text | Function callingTool choice | 131.1K | Same provider Overall 83% Missing image |
| NVIDIA-Nemotron-Nano-9B-v2 deepinfra | In $0.0400 / 1M tokens Out $0.1600 / 1M tokens | imagetext
Output: text | Function callingTool choice | 131.1K | Same provider Overall 83% Missing image |
| Meta-Llama-3.1-8B-Instruct deepinfra | In $0.0300 / 1M tokens Out $0.0500 / 1M tokens | imagetext
Output: text | Function callingTool choice | 131.1K | Same provider Overall 82% Missing image |
| Llama-4-Scout-17B-16E-Instruct deepinfra | In $0.0800 / 1M tokens Out $0.3000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 327.7K | Same provider Overall 78% |
| claude-3-7-sonnet-latest deepinfra | In $3.3000 / 1M tokens Out $16.5000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 200.0K | Same provider Overall 75% |
| claude-4-sonnet deepinfra | In $3.3000 / 1M tokens Out $16.5000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 200.0K | Same provider Overall 75% |
| Mistral-Small-24B-Instruct-2501 deepinfra | In $0.0500 / 1M tokens Out $0.0800 / 1M tokens | imagetext
Output: text | Function callingTool choice | 32.8K | Same provider Overall 75% Missing image |
| claude-4-opus deepinfra | In $16.5000 / 1M tokens Out $82.5000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 200.0K | Same provider Overall 75% |
| Llama-4-Maverick-17B-128E-Instruct-FP8 deepinfra | In $0.1500 / 1M tokens Out $0.6000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 1.0M | Same provider Overall 71% |
| mistral-medium-2505 azure_ai | In $0.4000 / 1M tokens Out $2.0000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 8.2K | Text + image covered Overall 67% |
| gpt-4 Azure | In $30.0000 / 1M tokens Out $60.0000 / 1M tokens | imagetext
Output: text | Function callingTool choice | 4.1K | Text + image covered Overall 65% |
| gemini-2.0-flash-001 deepinfra | In $0.1000 / 1M tokens Out $0.4000 / 1M tokens | imagetext audiocode
Output: text | Function callingTool choice | 1.0M | Same provider Overall 58% |
| google.gemma-3-4b-it bedrock_converse | In $0.0400 / 1M tokens Out $0.0800 / 1M tokens | imagetext
Output: image, text | Low overlap | 8.2K | Text + image covered Overall 53% |
| llama-3.1-8b-instant Groq | In $0.0500 / 1M tokens Out $0.0800 / 1M tokens | imagetext
Output: unknown | Function callingTool choice | 8.2K | Partial I/O overlap Overall 44% Missing text, image |
| gemma-7b-it Groq | In $0.0500 / 1M tokens Out $0.0800 / 1M tokens | imagetext
Output: unknown | Function callingTool choice | 8.2K | Partial I/O overlap Overall 44% Missing text, image |
| llama-3.2-3b llamagate | In $0.0400 / 1M tokens Out $0.0800 / 1M tokens | imagetext
Output: unknown | Function calling | 8.2K | Partial I/O overlap Overall 29% Missing text, image |
| Meta-Llama-3.2-1B-Instruct sambanova | In $0.0400 / 1M tokens Out $0.0800 / 1M tokens | imagetext
Output: unknown | Low overlap | 16.4K | Partial I/O overlap Overall 22% Missing text, image |
No models match this filter.