Model comparison

Gemma 4 12B vs GPT-5.6 Luna

Updated August 10, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.

Gemma 4 12B

Google

46.3/100

Estimated · Public rank #148

90% interval 34.8–57.8

GPT-5.6 Luna

OpenAI

66.9/100

Estimated · Public rank #23

90% interval 56.4–77.3

GPT-5.6 Luna has the higher public score estimate, 66.87 versus 46.31, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.

3 results are shared. Category rows based on different benchmark sets are marked directional and do not name a winner.

Which one for your work

Recommendations appear only when a shared evidence basis or an explicit operating constraint supports the call. Secondary and unsupported use cases stay disclosed below the initial list.

Long documents
Prompts that approach the documented context limit
GPT-5.6 Luna
GPT-5.6 Luna has the larger documented context window.
Confidence: documented

Show secondary and unsupported calls

Coding work
Code generation, repair, and software-engineering tasks
Not enough matched evidence
No shared weighted benchmark basis supports a winner.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
No shared weighted benchmark basis supports a winner.
Confidence: limited
Chat turn cost
1K fresh input + 500 output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Cache-heavy agent loop cost
200K cached + 20K fresh input + 10K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates
Repository review cost
50K fresh input + 3K output tokens
Not enough matched evidence
A complete comparable API-rate estimate is not available for both models.
Confidence: listed-rates

What is actually comparable

Shared results can support a head-to-head reading. Results present for only one model describe coverage, not superiority.

9 Gemma 4 12B only

3 Shared

19 GPT-5.6 Luna only

Shared results: 3
Gemma 4 12B only: 9
GPT-5.6 Luna only: 19
Like-for-like categories: 1 / 8

1 category uses different evidence sets. Those rows remain visible for coverage context but do not name a winner.

Category results, on a stated basis

Each row states whether both averages use the same weighted benchmark set. Directional and not-comparable rows remain visible, but they never receive a winner in this template.

Multimodal

Like-for-like

Gemma 4 12B: 69.1
GPT-5.6 Luna: 78.4
Weighted basis: 1 vs 1 rows
Reading: GPT-5.6 Luna leads

Knowledge

Directional only

Gemma 4 12B: 77.5
GPT-5.6 Luna: 92.3
Weighted basis: 2 vs 1 rows
Reading: Directional only

Agentic

Not comparable

Gemma 4 12B: Not measured
GPT-5.6 Luna: 84.1
Weighted basis: 0 vs 2 rows
Reading: Not comparable

Coding

Not comparable

Gemma 4 12B: Not measured
GPT-5.6 Luna: 62.7
Weighted basis: 0 vs 1 rows
Reading: Not comparable

Reasoning

Not comparable

Gemma 4 12B: 43.4
GPT-5.6 Luna: 59.5
Weighted basis: 1 vs 1 rows
Reading: Not comparable

Math

Not comparable

Gemma 4 12B: 77.5
GPT-5.6 Luna: 73.6
Weighted basis: 1 vs 2 rows
Reading: Not comparable

Multilingual

Not comparable

Gemma 4 12B: Not measured
GPT-5.6 Luna: Not measured
Weighted basis: 0 vs 0 rows
Reading: Not comparable

Instruction following

Not comparable

Gemma 4 12B: Not measured
GPT-5.6 Luna: Not measured
Weighted basis: 0 vs 0 rows
Reading: Not comparable

Category averages with the server-provided evidence basis for Gemma 4 12B and GPT-5.6 Luna
Category	Gemma 4 12B	GPT-5.6 Luna	Weighted basis	Reading
Multimodal	69.1	78.4	Like-for-like1 vs 1 rows	GPT-5.6 Luna leads
Knowledge	77.5	92.3	Directional only2 vs 1 rows	Directional only
Agentic	Not measured	84.1	Not comparable0 vs 2 rows	Not comparable
Coding	Not measured	62.7	Not comparable0 vs 1 rows	Not comparable
Reasoning	43.4	59.5	Not comparable1 vs 1 rows	Not comparable
Math	77.5	73.6	Not comparable1 vs 2 rows	Not comparable
Multilingual	Not measured	Not measured	Not comparable0 vs 0 rows	Not comparable
Instruction following	Not measured	Not measured	Not comparable0 vs 0 rows	Not comparable

Shape of the matched evidence

Only shared public evidence is shown. Sparse evidence stays a ruled list rather than being closed into a radar shape.

Too few matched category axes support a radar. The ruled list below shows only shared benchmark results; positions use each benchmark’s normalized display scale when available.

GPQA
Knowledge
Gemma 4 12B: 78.8%GPT-5.6 Luna: 92.3%Normalized gap 13.5Gemma 4 12B source GPT-5.6 Luna source
MMMU-Pro
Multimodal
Gemma 4 12B: 69.1%GPT-5.6 Luna: 78.4%Normalized gap 9.3Gemma 4 12B source GPT-5.6 Luna source

What each workload costs

Three fixed token mixes turn per-token rates into comparable decisions. Each scenario states context fit and whether cached input had to fall back to the published list-input rate.

Chat turn

1K fresh input + 500 output tokens

Gemma 4 12B: API rate not published; Fits in one request
GPT-5.6 Luna: $0.0008; Fits in one request

Gemma 4 12B has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

Gemma 4 12B: API rate not published; Fits in one request
GPT-5.6 Luna: $0.0136; Fits in one request

Gemma 4 12B has no comparable published API token rate.

Cache-heavy agent loop

200K cached + 20K fresh input + 10K output tokens

Gemma 4 12B: API rate not published; Fits in one request; Cached-input rate unavailable
GPT-5.6 Luna: $0.02; Fits in one request

Gemma 4 12B has no comparable published API token rate.

Specification differences

Sourced differences are shown directly. Missing facts stay explicit instead of being inferred from a model name or family.

SpecificationGemma 4 12BGPT-5.6 Luna

Context window

Maximum documented context; output-token limits may be lower.

Gemma 4 12B

256K

Google Gemma 4 model documentation

GPT-5.6 Luna

1.05M

OpenAI model catalog

Gemma 4 12B

Reasoning

GPT-5.6 Luna

Reasoning

Weight access

Gemma 4 12B

Open Weight

GPT-5.6 Luna

Proprietary

License

Gemma 4 12B

Open Weight

GPT-5.6 Luna

Proprietary

Release date

Gemma 4 12B

2026-06-03

GPT-5.6 Luna

2026-07-09

If you already use one of these models

Deployment change: The models list different providers, so authentication, endpoint behavior, limits, and feature support may change.
Quality signal: GPT-5.6 Luna has the higher public score estimate, 66.87 versus 46.31, but the 90% score intervals overlap.
Workload cost: A complete comparable API-rate estimate is not available for both models.
Context tradeoff: GPT-5.6 Luna has the larger documented window (1.05M).

Run the same representative tasks against both endpoints before changing production traffic.

Benchmark evidence

The full public result ledger is available for audit without forcing a wide desktop table onto a phone.

Browse raw public benchmark evidence31 rows

Agentic

Terminal-Bench 2.0
Gemma 4 12B—
GPT-5.6 Luna84.7%
Source
Not directly comparable
BrowseComp
Gemma 4 12B—
GPT-5.6 Luna83.3%
Source
Not directly comparable
OSWorld 2.0
Gemma 4 12B—
GPT-5.6 Luna45.6%
Source
Not directly comparable
CyberGym
Gemma 4 12B—
GPT-5.6 Luna77.9%
Source
Not directly comparable
ExploitGym
Gemma 4 12B—
GPT-5.6 Luna12.4%
Source
Not directly comparable
Toolathlon
Gemma 4 12B—
GPT-5.6 Luna53.4%
Source
Not directly comparable

Coding

LiveCodeBench v6
Gemma 4 12B72.0%
Source
GPT-5.6 Luna—
Not directly comparable
SWE-bench Pro
Gemma 4 12B—
GPT-5.6 Luna62.7%
Source
Not directly comparable
Terminal-Bench 2.0
Gemma 4 12B—
GPT-5.6 Luna84.7%
Source
Not directly comparable
deepSwe
Gemma 4 12B—
GPT-5.6 Luna67.2%
Source
Not directly comparable
FrontierCode 1.1 Extended
Gemma 4 12B—
GPT-5.6 Luna55.1%
Source
Not directly comparable
cursorBench32
Gemma 4 12B—
GPT-5.6 Luna61.1%
Source
Not directly comparable

Reasoning

BBH
Gemma 4 12B53%
Source
GPT-5.6 Luna—
Not directly comparable
MRCRv2
Gemma 4 12B43.4%
Source
GPT-5.6 Luna—
Not directly comparable
ARC-AGI-2
Gemma 4 12B—
GPT-5.6 Luna59.5%
Source
Not directly comparable
ARC-AGI-3
Gemma 4 12B—
GPT-5.6 Luna0.2%
Source
Not directly comparable

Knowledge

GPQA
Gemma 4 12B78.8%
Source
GPT-5.6 Luna92.3%
Source
GPT-5.6 Luna leads this result
GPQA-D
Gemma 4 12B78.8%
Source
GPT-5.6 Luna92.3%
Source
GPT-5.6 Luna leads this result
MMLU-Pro
Gemma 4 12B77.2%
Source
GPT-5.6 Luna—
Not directly comparable
HLE w/o tools
Gemma 4 12B5.2%
Source
GPT-5.6 Luna—
Not directly comparable
MMMLU
Gemma 4 12B83.4%
Source
GPT-5.6 Luna—
Not directly comparable
HealthBench Professional
Gemma 4 12B—
GPT-5.6 Luna55.7%
Source
Not directly comparable
HealthBench Hard
Gemma 4 12B—
GPT-5.6 Luna32.0%
Source
Not directly comparable

Math

AIME26
Gemma 4 12B77.5%
Source
GPT-5.6 Luna—
Not directly comparable
FrontierMath (legacy)
Gemma 4 12B—
GPT-5.6 Luna78.6%
Source
Not directly comparable
FrontierMath v2 (Tiers 1-3)
Gemma 4 12B—
GPT-5.6 Luna78.600%
Source
Not directly comparable
FrontierMath v2 (Tier 4)
Gemma 4 12B—
GPT-5.6 Luna58.500%
Source
Not directly comparable

Multimodal

MMMU-Pro
Gemma 4 12B69.1%
Source
GPT-5.6 Luna78.4%
Source
GPT-5.6 Luna leads this result
MathVision
Gemma 4 12B79.7%
Source
GPT-5.6 Luna—
Not directly comparable
MedXpertQA (MM)
Gemma 4 12B48.7%
Source
GPT-5.6 Luna—
Not directly comparable
MMMU-Pro w/ Python
Gemma 4 12B—
GPT-5.6 Luna79.5%
Source
Not directly comparable

Frequently asked questions

Which is better, Gemma 4 12B or GPT-5.6 Luna?

GPT-5.6 Luna has the higher public score estimate, 66.87 versus 46.31, but the 90% score intervals overlap. The higher estimate is not a decisive winner because the uncertainty ranges overlap.

Which is better for coding, Gemma 4 12B or GPT-5.6 Luna?

The published evidence does not provide a shared weighted coding basis for both models, so BenchLM does not name a coding winner.

Which is better for agentic tasks, Gemma 4 12B or GPT-5.6 Luna?

The published evidence does not provide a shared weighted agentic tasks basis for both models, so BenchLM does not name a agentic tasks winner.

Which costs less, Gemma 4 12B or GPT-5.6 Luna?

Both models do not have comparable published API token rates, so this page does not name a universal price winner.

Which has the larger context window, Gemma 4 12B or GPT-5.6 Luna?

GPT-5.6 Luna has the larger documented context window: 1.05M, compared with 256K.

Related comparisons

Compare API pricing Read the methodology Open the model selector

Last updated August 10, 2026

Watch Gemma 4 12B vs GPT-5.6 Luna

One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.

Read a sample issue

Join 2,000+ readers.

Gemma 4 12B vs GPT-5.6 Luna

Which one for your work

Long documents

Coding work

Agentic work

Chat turn cost

Cache-heavy agent loop cost

Repository review cost

What is actually comparable

Category results, on a stated basis

Multimodal

Knowledge

Agentic

Coding

Reasoning

Math

Multilingual

Instruction following

Shape of the matched evidence

What each workload costs

Chat turn

Repository review

Cache-heavy agent loop

Specification differences

Context window

API model ID

Cached-input rate

Documented inputs

Documented outputs

Provider availability

Reasoning profile

Weight access

License

Release date

Benchmark evidence

Agentic

Coding

Reasoning

Knowledge

Math

Multimodal

Frequently asked questions

Which is better, Gemma 4 12B or GPT-5.6 Luna?

Which is better for coding, Gemma 4 12B or GPT-5.6 Luna?

Which is better for agentic tasks, Gemma 4 12B or GPT-5.6 Luna?

Which costs less, Gemma 4 12B or GPT-5.6 Luna?

Which has the larger context window, Gemma 4 12B or GPT-5.6 Luna?

Related comparisons

Watch Gemma 4 12B vs GPT-5.6 Luna