Gemma 4 31B vs GPT-5.5

Updated November 13, 2026. Rank says GPT-5.5 is ahead. Price, access, and your workload can each overturn that. Public scores include evidence status and uncertainty.

Decision reading

GPT-5.5 has the higher public point estimate, 66.01 versus 28.81. Their conditional score ranges do not overlap. These ranges do not establish rank confidence. 7 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.

Model A

Gemma 4 31B

Google

28.81/100

Estimated · Public rank #175

Conditional range 19.1–38.5

Model B

GPT-5.5

OpenAI

66.01/100

Supported · Public rank #23

90% interval 59.0–73.1

API price per 1M tokens

Gemma 4 31B: No comparable API price

GPT-5.5: $5 input / $30 output

Context window

Gemma 4 31B: 256K

GPT-5.5: 1M

Released

Gemma 4 31B: April 2, 2026

GPT-5.5: April 23, 2026

Shared results: 7
Gemma 4 31B only: 6
GPT-5.5 only: 43
Like-for-like categories: 4 / 8

Estimated: Gemma 4 31B · Supported: GPT-5.5. Conditional ranges do not establish rank confidence.How the comparison works

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.

Coding work
Code generation, repair, and software-engineering tasks
GPT-5.5
GPT-5.5 has the higher public coding point estimate, 57.7 to 28.8, with Supported evidence for both models. The conditional ranges do not establish rank confidence.
Confidence: limited
Long documents
Prompts that approach the documented context limit
GPT-5.5
GPT-5.5 has the larger documented context window.
Confidence: documented

Show secondary and unsupported calls

Agentic work
Tool use, computer use, and multi-step task completion
Not enough matched evidence
Gemma 4 31B is scored on Estimated evidence for agentic, so the reading is directional rather than like-for-like.
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

Which one for a specific job

Choose a job from the LLM Selector's task catalog to see the category row it rests on, under the same basis rules as the table below. A directional row stays directional; choosing a job never creates a winner.

Pick the job you would use them for

The same task catalog as the LLM Selector. Each job names the evidence surface it rests on; nothing here adds a new score.

28.8Gemma 4 31B57.7GPT-5.5

Like-for-like · BenchAlign v5.8

GPT-5.5 has the higher coding point estimate. Conditional score ranges do not establish rank confidence.

Coding scores combine specific tasks and setups. Match the editor, harness, and effort to your workflow.

Open the coding leaderboard Shortlist models for this job

Same basis rules as the category table below

Or describe your own job

The LLM Selector reads what you need a model for and returns a shortlist, ranked in the same order the job pages publish.

What do you need a model for?

Describe the task and constraints. This selector recommends models; it cannot carry out the task or access your files and accounts.

We keep what you type here, drafts included, for 180 days, and after that only the job and requirements read from it. Nothing is tied to your IP address or account. Leave out names and other personal details. Privacy

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.

HLEKnowledge
Normalized gap 25.7
Gemma 4 31B:26.5%
GPT-5.5:52.2%
Gemma 4 31B source GPT-5.5 source
HLE w/o toolsKnowledge
Normalized gap 21.9
Gemma 4 31B:19.5%
GPT-5.5:41.4%
Gemma 4 31B source GPT-5.5 source
GPQAKnowledge
Normalized gap 9.3
Gemma 4 31B:84.3%
GPT-5.5:93.6%
Gemma 4 31B source GPT-5.5 source
MMMU-ProMultimodal
Normalized gap 4.3
Gemma 4 31B:76.9%
GPT-5.5:81.2%
Gemma 4 31B source GPT-5.5 source

Bars run 0–100 on each benchmark’s normalized display scale

Category results, on a stated basis

Each row shows the public-lane category score for both models: the BenchAlign v5.8 lane for agentic, coding, reasoning, multimodal, and knowledge, and the weighted public rows elsewhere, with the same rank each model page reports. A row is like-for-like only when both scores rest on Supported evidence or the same weighted set. Directional and not-comparable rows remain visible, but they never receive a winner.

Coding

Like-for-like

Gemma 4 31B: 28.8; Supported · #107/162
GPT-5.5: 57.7; Supported · #20/162
Basis: BenchAlign v5.8 lane · 3 vs 14 public rows
Reading: GPT-5.5 leads

Reasoning

Like-for-like

Gemma 4 31B: 32.0; Supported · #109/160
GPT-5.5: 65.8; Supported · #20/160
Basis: BenchAlign v5.8 lane · 1 vs 6 public rows
Reading: GPT-5.5 leads

Multimodal

Like-for-like

Gemma 4 31B: 47.3; Supported · #60/67
GPT-5.5: 67.3; Supported · #17/67
Basis: BenchAlign v5.8 lane · 2 vs 5 public rows
Reading: GPT-5.5 leads

Knowledge

Like-for-like

Gemma 4 31B: 36.7; Supported · #121/180
GPT-5.5: 69.3; Supported · #16/180
Basis: BenchAlign v5.8 lane · 4 vs 6 public rows
Reading: GPT-5.5 leads

Agentic

Directional only

Gemma 4 31B: 12.5; Estimated · #133/135
GPT-5.5: 57.8; Supported · #32/135
Basis: BenchAlign v5.8 lane · 3 vs 16 public rows
Reading: Directional only

Instruction following

Directional only

Gemma 4 31B: 92.9; #16/157
GPT-5.5: 93.3; #9/157
Basis: Provisional lane · 0 vs 0 weighted rows
Reading: Directional only

Multilingual

Not comparable

Gemma 4 31B: Not ranked
GPT-5.5: Not ranked
Basis: Provisional lane · 0 vs 0 weighted rows
Reading: Not comparable

Math

Not comparable

Gemma 4 31B: Not ranked
GPT-5.5: 85.9; Unranked · 3 rankable rows
Basis: Provisional lane · 0 vs 2 weighted rows
Reading: Not comparable

Category scores with the server-provided evidence basis for Gemma 4 31B and GPT-5.5
Category	Gemma 4 31B	GPT-5.5	Basis	Reading
Coding	28.8Supported · #107/162	57.7Supported · #20/162	Like-for-likeBenchAlign v5.8 lane · 3 vs 14 public rows	GPT-5.5 leads
Reasoning	32.0Supported · #109/160	65.8Supported · #20/160	Like-for-likeBenchAlign v5.8 lane · 1 vs 6 public rows	GPT-5.5 leads
Multimodal	47.3Supported · #60/67	67.3Supported · #17/67	Like-for-likeBenchAlign v5.8 lane · 2 vs 5 public rows	GPT-5.5 leads
Knowledge	36.7Supported · #121/180	69.3Supported · #16/180	Like-for-likeBenchAlign v5.8 lane · 4 vs 6 public rows	GPT-5.5 leads
Agentic	12.5Estimated · #133/135	57.8Supported · #32/135	Directional onlyBenchAlign v5.8 lane · 3 vs 16 public rows	Directional only
Instruction following	92.9#16/157	93.3#9/157	Directional onlyProvisional lane · 0 vs 0 weighted rows	Directional only
Multilingual	Not ranked	Not ranked	Not comparableProvisional lane · 0 vs 0 weighted rows	Not comparable
Math	Not ranked	85.9Unranked · 3 rankable rows	Not comparableProvisional lane · 0 vs 2 weighted rows	Not comparable

Ranks count the models scored in each category’s lane, so the agentic, coding, reasoning, multimodal, and knowledge denominators (BenchAlign v5.8) differ from the provisional-lane categories. Unranked scores sit on the provisional lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.

Supported evidence per lane · bars run 0–100Methodology

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 31B: Self-hosted; infrastructure cost varies; Fits in one request
GPT-5.5: $0.02; Fits in one request

Gemma 4 31B has no comparable published API token rate.

Repository review

50K fresh input + 3K output tokens

Gemma 4 31B: Self-hosted; infrastructure cost varies; Fits in one request
GPT-5.5: $0.34; Fits in one request

Gemma 4 31B has no comparable published API token rate.

Cache-heavy agent loop

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

Gemma 4 31B: Self-hosted; infrastructure cost varies; Fits in one request; Cached-input rate unavailable
GPT-5.5: $0.5; Fits in one request

Gemma 4 31B has no comparable published API token rate.

Cached input falls back to the list input rate only where a cached rate is unpublished

Price against score, among all priced models

Overall score against output price per 1M tokens, one model per family. Pricing as of November 12, 2026.

GPT-5.5

Gemma 4 31B has no public API price, so it is not plotted.
11 cheaper models score at least as high as GPT-5.5. The nearest is Claude Opus 5 (score 76.1, $25.00 per 1M output tokens).

Gray dots are the other priced models. The dark step line is the efficiency frontier: for a model under it, another model costs no more and scores at least as high.

See GPT-5.5 on the full price vs performance chart

GPT-5.5

Not sourced

Reasoning profile

Gemma 4 31B

Reasoning

GPT-5.5

Reasoning

Weight access

Gemma 4 31B

Open Weight

GPT-5.5

Proprietary

License

Gemma 4 31B

Open Weight

GPT-5.5

Proprietary

Release date

Gemma 4 31B

2026-04-02

GPT-5.5

2026-04-23

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-5.5 has the higher public point estimate, 66.01 versus 28.81. Their conditional score ranges do not overlap. These ranges do not establish rank confidence.
Workload cost: A complete comparable API-rate estimate is not available for both models.
Context tradeoff: GPT-5.5 has the larger documented window (1M).

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

Questions

Which is better, Gemma 4 31B or GPT-5.5?

GPT-5.5 has the higher public point estimate, 66.01 versus 28.81. Their conditional score ranges do not overlap. These ranges do not establish rank confidence. The page therefore keeps the decision tied to the specific documented workload.

Which is better for coding, Gemma 4 31B or GPT-5.5?

GPT-5.5 has the higher public coding point estimate, 57.7 to 28.8, with Supported evidence for both models. The conditional ranges do not establish rank confidence.

Which is better for agentic tasks, Gemma 4 31B or GPT-5.5?

GPT-5.5 scores higher for agentic tasks on the public lane, 57.8 to 12.5. Gemma 4 31B is scored on Estimated evidence for agentic tasks, so the reading is directional rather than like-for-like. BenchLM does not name a winner for agentic tasks from a directional reading; read the shared benchmark rows directly and test the models on the same task set.

Which costs less, Gemma 4 31B or GPT-5.5?

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 31B or GPT-5.5?

GPT-5.5 has the larger documented context window: 1M, compared with 256K.

Self-host vs API cost

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

Gemma 4 31B

API / mo (Novita AI)$405

Self-host / mo$298

Break-even37M/day

GPT-5.5

API / mo$26,250

Self-host / moNot listed

Break-even—

Proprietary model — self-hosting not applicable.

Model the full break-even

Benchmark evidence

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

Browse raw public benchmark evidence56 rows

Agentic

b3 Agent Security
Gemma 4 31B83.0%
Source
GPT-5.5—
Not directly comparable
Cyber Misuse Refusal
Gemma 4 31B84.0%
Source
GPT-5.5—
Not directly comparable
Gert Labs
Shared source
Gemma 4 31B35.26%
GPT-5.572.93%
GPT-5.5 leads this result
LifeVal
Gemma 4 31B—
GPT-5.577.4%
Source
Not directly comparable
Terminal-Bench 2.0
Gemma 4 31B—
GPT-5.582%
Source
Not directly comparable
CyberGym
Gemma 4 31B—
GPT-5.581.8%
Source
Not directly comparable
BrowseComp
Gemma 4 31B—
GPT-5.584.4%
Source
Not directly comparable
OSWorld-Verified
Gemma 4 31B—
GPT-5.578.7%
Source
Not directly comparable
MCP Atlas
Gemma 4 31B—
GPT-5.575.3%
Source
Not directly comparable
Toolathlon
Gemma 4 31B—
GPT-5.555.6%
Source
Not directly comparable
τ²-bench results
Gemma 4 31B—
GPT-5.598%
Source
Not directly comparable
ResearchClawBench
Gemma 4 31B—
GPT-5.517.0%
Source
Not directly comparable
OSWorld 2.0
Gemma 4 31B—
GPT-5.513.0%
Source
Not directly comparable
JobBench
Gemma 4 31B—
GPT-5.542.7%
Source
Not directly comparable
ExploitGym
Gemma 4 31B—
GPT-5.513.4%
Source
Not directly comparable
Terminal-Bench 2.1 (Vals)
Gemma 4 31B—
GPT-5.576.4%
Source
Not directly comparable
ApprenticeBench
Gemma 4 31B—
GPT-5.520%
Source
Not directly comparable
MazeBench
Gemma 4 31B—
GPT-5.50%
Source
Not directly comparable

Coding

LiveCodeBench v6
Gemma 4 31B80.0%
Source
GPT-5.5—
Not directly comparable
SWE-Rebench
Gemma 4 31B41.6%
Source
GPT-5.5—
Not directly comparable
React Native Evals
Shared source
Gemma 4 31B75.2%
GPT-5.584.7%
GPT-5.5 leads this result
SWE-bench Pro
Gemma 4 31B—
GPT-5.558.6%
Source
Not directly comparable
Terminal-Bench 2.0
Gemma 4 31B—
GPT-5.582.0%
Source
Not directly comparable
Vibe Code Bench
Gemma 4 31B—
GPT-5.569.85%
Source
Not directly comparable
cursorBench31
Gemma 4 31B—
GPT-5.559.2%
Source
Not directly comparable
CursorBench 3.2
Gemma 4 31B—
GPT-5.558.4%
Source
Not directly comparable
FrontierCode 1.1 Main
Gemma 4 31B—
GPT-5.543.0%
Source
Not directly comparable
LiveCodeBench (Vals)
Gemma 4 31B—
GPT-5.585.3%
Source
Not directly comparable
SWE-bench (Vals)
Gemma 4 31B—
GPT-5.582.6%
Source
Not directly comparable
PostTrainBench v1.1
Gemma 4 31B—
GPT-5.527.2%
Source
Not directly comparable
FrontierCode 2.0
Gemma 4 31B—
GPT-5.522.6%
Source
Not directly comparable
VulcanBench Frontier v4
Gemma 4 31B—
GPT-5.578.5%
Source
Not directly comparable
PostTrainBench
Gemma 4 31B—
GPT-5.523.9%
Source
Not directly comparable
PostTrainBench v1
Gemma 4 31B—
GPT-5.525.0%
Source
Not directly comparable

Reasoning

MRCRv2
Gemma 4 31B66.4%
Source
GPT-5.5—
Not directly comparable
BioBasicsBench
Gemma 4 31B—
GPT-5.535.4%
Source
Not directly comparable
MRCR v2 64K-128K
Gemma 4 31B—
GPT-5.583.1%
Source
Not directly comparable
MRCR v2 128K-256K
Gemma 4 31B—
GPT-5.587.5%
Source
Not directly comparable
ARC-AGI-2
Gemma 4 31B—
GPT-5.585%
Source
Not directly comparable
ARC-AGI-3
Gemma 4 31B—
GPT-5.50.4%
Source
Not directly comparable
ARC-AGI-1
Gemma 4 31B—
GPT-5.595.00%
Source
Not directly comparable

Multimodal

ParseBench-1
Shared source
Gemma 4 31B62.40%
GPT-5.566.42%
GPT-5.5 leads this result
MMMU-Pro
Gemma 4 31B76.9%
Source
GPT-5.581.2%
Source
GPT-5.5 leads this result
ExtractBench (LlamaIndex)
Gemma 4 31B—
GPT-5.593.35%
Source
Not directly comparable
MMMU-Pro w/ Python
Gemma 4 31B—
GPT-5.583.2%
Source
Not directly comparable
OfficeQA Pro
Gemma 4 31B—
GPT-5.554.1%
Source
Not directly comparable

Knowledge

GPQA
Gemma 4 31B84.3%
Source
GPT-5.593.6%
Source
GPT-5.5 leads this result
MMLU-Pro
Gemma 4 31B85.2%
Source
GPT-5.5—
Not directly comparable
HLE
Gemma 4 31B26.5%
Source
GPT-5.552.2%
Source
GPT-5.5 leads this result
HLE w/o tools
Gemma 4 31B19.5%
Source
GPT-5.541.4%
Source
GPT-5.5 leads this result
GPQA-D
Gemma 4 31B—
GPT-5.593.6%
Source
Not directly comparable
GPQA Diamond (Vals)
Gemma 4 31B—
GPT-5.593.2%
Source
Not directly comparable
MMLU-Pro (Vals)
Gemma 4 31B—
GPT-5.588.1%
Source
Not directly comparable

Math

FrontierMath (legacy)
Gemma 4 31B—
GPT-5.551.7%
Source
Not directly comparable
FrontierMath v2 (Tiers 1-3)
Gemma 4 31B—
GPT-5.585.263%
Source
Not directly comparable
FrontierMath v2 (Tier 4)
Gemma 4 31B—
GPT-5.572.500%
Source
Not directly comparable

56 public results · 7 shared

Related comparisons

Compare API pricing Read the methodology Open the model selector

Watch Gemma 4 31B vs GPT-5.5

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

Read a sample issue

Join 7,000+ subscribers.

Last updated November 13, 2026

Gemma 4 31B vs GPT-5.5

Decision reading

Which one for your work

Coding work

Long documents

Agentic work

Chat turn cost

Cache-heavy agent loop cost

Repository review cost

Which one for a specific job

Or describe your own job

What is actually comparable

Shape of the matched evidence

Category results, on a stated basis

Coding

Reasoning

Multimodal

Knowledge

Agentic

Instruction following

Multilingual

Math

What each workload costs

Chat turn

Repository review

Cache-heavy agent loop

Price against score, among all priced models

Specification differences

Context window

API model ID

Cached-input rate

Documented inputs

Documented outputs

Provider availability

Reasoning profile

Weight access

License

Release date

If you already use one of these models

Questions

Which is better, Gemma 4 31B or GPT-5.5?

Which is better for coding, Gemma 4 31B or GPT-5.5?

Which is better for agentic tasks, Gemma 4 31B or GPT-5.5?

Which costs less, Gemma 4 31B or GPT-5.5?

Which has the larger context window, Gemma 4 31B or GPT-5.5?

Self-host vs API cost

Benchmark evidence

Agentic

Coding

Reasoning

Multimodal

Knowledge

Math

Related comparisons

Watch Gemma 4 31B vs GPT-5.5