Skip to main content
BenchLM
Advertise

Model comparison

Gemma 4 26B A4B vs Gemma 4 31B

Data verified

Head-to-head evidence from 20 shared benchmark results across 6 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

Sibling matchup inside the Gemma 4 family.

57.96/100
Margin
3.1pts
winning →
61.08/100
0 category wins2 category wins

Public leaderboard positions: Gemma 4 26B A4B #67 (Supported); Gemma 4 31B #43 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Gemma 4 26B A4B and Gemma 4 31B share 20 comparable benchmark results. 2 of 8 categories are comparable. 0 results are unique to Gemma 4 26B A4B; 9 to Gemma 4 31B.

Updated July 23, 2026
Shared results
20
Gemma 4 26B A4B only
0
Gemma 4 31B only
9
Comparable categories
2 / 8

Gemma 4 26B A4B makes more sense if you want this variant’s specific profile, while Gemma 4 31B is the cleaner fit if knowledge is the priority.

Confidence note. This is a partial-evidence comparison with 20 shared benchmark results across 6 evidence categories; 2 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

Gemma 4 26B A4B and Gemma 4 31B sit in the same Gemma 4 family. This page is less about two unrelated model lineages and more about how the siblings trade off on benchmark shape, token costs, and practical limits like context window.

Gemma 4 31B is clearly ahead on the BenchAlign aggregate, 61.08 to 57.96. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

Gemma 4 31B's sharpest advantage is in knowledge, where it averages 52.9 against 43.4. The single biggest benchmark swing on the page is HLE, 17.2% to 26.5%.

Category breakdown

Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.

Category scores and score margins for Gemma 4 26B A4B and Gemma 4 31B
CategoryGemma 4 26B A4BΔGemma 4 31B
KnowledgeGemma 4 26B A4B43.4Margin→ 9.5Gemma 4 31B52.9
MultimodalGemma 4 26B A4B73.8Margin→ 3.1Gemma 4 31B76.9
CodingGemma 4 26B A4BNot measuredMarginNo overlapGemma 4 31B41.6

Decisive benchmark drivers

The largest measured benchmark gaps in this matchup, with exact reported values.

More
A · Gemma 4 26B A4BB · Gemma 4 31B

  1. HLE

    Knowledge
    Source ↗
    A 17.2%B 26.5%
    Winner: Gemma 4 31BΔ 9.3
    HLE: Gemma 4 26B A4B scored 17.2%; Gemma 4 31B scored 26.5%. Gemma 4 31B wins this benchmark.

  2. MMMU-Pro

    Multimodal
    Source ↗
    A 73.8%B 76.9%
    Winner: Gemma 4 31BΔ 3.1
    MMMU-Pro: Gemma 4 26B A4B scored 73.8%; Gemma 4 31B scored 76.9%. Gemma 4 31B wins this benchmark.

  3. MMLU-Pro

    Knowledge
    Source ↗
    A 82.6%B 85.2%
    Winner: Gemma 4 31BΔ 2.6
    MMLU-Pro: Gemma 4 26B A4B scored 82.6%; Gemma 4 31B scored 85.2%. Gemma 4 31B wins this benchmark.

Operational comparison

Runtime and commercial metrics are compared only when both models have a complete sourced value.

MetricGemma 4 26B A4BGemma 4 31BComparison
Input / output priceUSD per 1M tokensGemma 4 26B A4B$0 input / $0 outputGemma 4 31B$0 input / $0 outputListed prices are equal.
Generation speedtokens per secondGemma 4 26B A4BNot availableGemma 4 31BNot availableA complete speed comparison is not available.
First-answer latencyseconds to first tokenGemma 4 26B A4BNot availableGemma 4 31BNot availableA complete latency comparison is not available.
Context windowmaximum listed tokensGemma 4 26B A4B256KGemma 4 31B256KListed context windows are equal.

Benchmark Deep Dive

Agentic
BenchmarkGemma 4 26B A4BGemma 4 31BResult
AA Agentic IndexSource 11.0%14.4%Gemma 4 31B leads
τ²-bench resultsSource 43.6%59.9%Gemma 4 31B leads
GDPval-AASource 13.1%15.2%Gemma 4 31B leads
GDPval-AASource 761804Gemma 4 31B leads
Gert LabsSource —35.26%Not comparable
AA EnterpriseOps-GymSource —28.3%Not comparable
AA ITBenchSource —37.3%Not comparable
AA Tau3 BankingSource —15.1%Not comparable
terminalBenchHardSource —36.4%Not comparable
Coding
BenchmarkGemma 4 26B A4BGemma 4 31BResult
AA Coding IndexSource 39.3%43.4%Gemma 4 31B leads
AA-SciCodeSource 40.0%43.4%Gemma 4 31B leads
SWE-RebenchSource —41.6%Not comparable
React Native EvalsSource —75.2%Not comparable
Reasoning
BenchmarkGemma 4 26B A4BGemma 4 31BResult
AA-LCRSource 55.7%62.0%Gemma 4 31B leads
CritPtSource 0.0%1.4%Gemma 4 31B leads
KnowledgeGemma 4 31B wins
BenchmarkGemma 4 26B A4BGemma 4 31BResult
MMLU-ProSource 82.6%85.2%Gemma 4 31B leads
HLESource 17.2%26.5%Gemma 4 31B leads
HLE w/o toolsSource 8.7%19.5%Gemma 4 31B leads
Artificial Analysis Intelligence IndexSource 25.7%29.4%Gemma 4 31B leads
AA-GPQA DiamondSource 79.2%85.7%Gemma 4 31B leads
AA-HLESource 18.3%22.7%Gemma 4 31B leads
AA-Omniscience IndexSource -48.1%-45.4%Gemma 4 31B leads
AA-Omniscience AccuracySource 18.2%19.9%Gemma 4 31B leads
AA-Omniscience Hallucination RateSource 80.9%81.6%Gemma 4 26B A4B leads
GPQASource —84.3%Not comparable
AA Openness IndexSource —38.9%Not comparable
MultimodalGemma 4 31B wins
BenchmarkGemma 4 26B A4BGemma 4 31BResult
MMMU-ProSource 73.8%76.9%Gemma 4 31B leads
AA-MMMU-ProSource 69.2%73.4%Gemma 4 31B leads
Inst. Following
BenchmarkGemma 4 26B A4BGemma 4 31BResult
AA-IFBenchSource 72.4%75.6%Gemma 4 31B leads
Frequently Asked Questions (3)

Which is better, Gemma 4 26B A4B or Gemma 4 31B?

Gemma 4 26B A4B and Gemma 4 31B are sibling variants in the Gemma 4 family, so the right pick depends on whether you value the better benchmark line, cheaper tokens, or the larger context window. Gemma 4 31B is ahead on BenchLM's BenchAlign leaderboard 61.08 to 57.96.

Which is better for knowledge tasks, Gemma 4 26B A4B or Gemma 4 31B?

Gemma 4 31B has the edge for knowledge tasks in this comparison, averaging 52.9 versus 43.4. Inside this category, HLE w/o tools is the benchmark that creates the most daylight between them.

Which is better for multimodal and grounded tasks, Gemma 4 26B A4B or Gemma 4 31B?

Gemma 4 31B has the edge for multimodal and grounded tasks in this comparison, averaging 76.9 versus 73.8. Inside this category, AA-MMMU-Pro is the benchmark that creates the most daylight between them.

Self-host vs API cost

Estimates at 50,000 req/day · 1000 tokens/req average.

Gemma 4 26B A4B
API / mo$0
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Gemma 4 31B
API / mo$0
Self-host / mo$429
Break-even—
Model the full break-even

Related Comparisons

Explore More

Google Compare Pricing Methodology Find Your Best LLM Overall Rankings
Last updated: July 23, 2026

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.