head-to-head

Google: Gemma 4 26B A4B (free) vs Google: Gemma 4 26B A4B

Side-by-side comparison of specs, pricing, benchmark scores, and task rankings. Updated 2026-11-07.

Google: Gemma 4 26B A4B (free) Google: Gemma 4 26B A4B
Vendorgooglegoogle
Quality Score100100
Benchmark Score48.448.4
Input PriceFree$0.07/M
Output PriceFree$0.23/M
Context Window262,144262,144
Max Output32,768235,929
Tool Calling✓✓
Structured Output✓✓
Reasoning Mode✓✓
Vision✓✓
Audio--
Benchmark Scores
ai_index42.442.4
ai_index_agentic18.118.1
ai_index_coding64.964.9
eqbench56.356.3

The verdict

For a sample job of 1 million input tokens and 250,000 output tokens, Gemma 4 26B A4B (free) costs $0 and Gemma 4 26B A4B costs $0.12, so Gemma 4 26B A4B (free) is free for the same work.

Both read up to 262,144 tokens at once, roughly 393 pages of text.

Across the 6 tasks where they differ, Gemma 4 26B A4B (free) scores higher on 6 (Code Completion, Regex Writing, JSON Extraction and others).

Pick Gemma 4 26B A4B (free) if you need lower cost or the stronger all-round task scores.

Costs use current list prices per million tokens; page counts assume about 0.75 words per token and 500 words per page.

Who wins by task?

TaskGoogle: Gemma 4 26B A4B (free)Google: Gemma 4 26B A4B
Code Completion 131 131
Regex Writing 130 130
JSON Extraction 142 142
Bulk Data Labeling 133 133
Short-Form Summarization 129 129
Blog Post Writing 132 131

Scores reflect capability match + benchmark data + pricing for each task. Methodology →

Related comparisons

Anthropic: Claude Haiku Latest vs Google: Gemma 4 26B A4B Anthropic: Claude Haiku Latest vs Google: Gemma 4 26B A4B (free) Anthropic: Claude Sonnet 5 vs Google: Gemma 4 26B A4B Anthropic: Claude Sonnet 5 vs Google: Gemma 4 26B A4B (free) Anthropic: Claude Sonnet Latest vs Google: Gemma 4 26B A4B Anthropic: Claude Sonnet Latest vs Google: Gemma 4 26B A4B (free) Google: Gemini 3.1 Flash Lite vs Google: Gemma 4 26B A4B Google: Gemini 3.1 Flash Lite vs Google: Gemma 4 26B A4B (free)

The Model Movers Report

One email every Friday, built from this site's own rankings: the current top five by benchmark score, every model released in the last seven days, and one note worked out from that week's numbers. You can unsubscribe from any issue with one click.