Agentic
Like-for-like- Gemma 4 31B
- 27.6
- Supported · #140/151
- GPT-5.2
- 43.0
- Supported · #108/151
- Basis
- BenchAlign lane · 1 vs 4 public rows
- Reading
- GPT-5.2 leads · intervals overlap
Gemma 4 31B vs GPT-5.2
Updated September 4, 2026. Public scores include evidence status and uncertainty. They are not guarantees for a specific workload.
Decision reading
GPT-5.2 has the higher public score estimate, 65.45 versus 58.69, but the 90% score intervals overlap. Treat that as a lead, not a settled winner.
3 results are shared. Category rows resting on Estimated evidence or different benchmark sets are marked directional and do not name a winner.
Share or export
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-5.2
GPT-5.2 leads on the public coding lane, 47.3 to 43.5, with Supported evidence for both models, although the 90% intervals overlap.
Confidence: limited
Agentic work
Tool use, computer use, and multi-step task completion
GPT-5.2
GPT-5.2 leads on the public agentic lane, 43 to 27.6, with Supported evidence for both models, although the 90% intervals overlap.
Confidence: limited
Long documents
Prompts that approach the documented context limit
GPT-5.2
GPT-5.2 has the larger documented context window.
Confidence: documented
Show secondary and unsupported calls
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: rate-fallback
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.
5 Gemma 4 31B only
3 Shared
12 GPT-5.2 only
- Shared results
- 3
- Gemma 4 31B only
- 5
- GPT-5.2 only
- 12
- Like-for-like categories
- 3 / 8
2 categories rest on Estimated evidence or different evidence sets. Those rows remain visible for coverage context but do not name a winner.
Category results, on a stated basis
Each row shows the public-lane category score for both models: the BenchAlign lane for agentic, coding, 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
- 43.5
- Supported · #123/183
- GPT-5.2
- 47.3
- Supported · #91/183
- Basis
- BenchAlign lane · 2 vs 3 public rows
- Reading
- GPT-5.2 leads · intervals overlap
Knowledge
Like-for-like- Gemma 4 31B
- 47.9
- Supported · #104/181
- GPT-5.2
- 62.0
- Supported · #30/181
- Basis
- BenchAlign lane · 4 vs 1 public rows
- Reading
- GPT-5.2 leads · intervals overlap
Multimodal
Directional only- Gemma 4 31B
- 58.4
- #30/48
- GPT-5.2
- 66.3
- #23/48
- Basis
- Provisional lane · 1 vs 2 weighted rows
- Reading
- Directional only
Instruction following
Directional only- Gemma 4 31B
- 92.6
- #12/120
- GPT-5.2
- 92.3
- #14/120
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Directional only
Reasoning
Not comparable- Gemma 4 31B
- 70.1
- Unranked · 2 rankable rows
- GPT-5.2
- 53.8
- Unranked · 3 rankable rows
- Basis
- Provisional lane · 0 vs 1 weighted rows
- Reading
- Not comparable
Math
Not comparable- Gemma 4 31B
- Not ranked
- GPT-5.2
- 57.5
- Unranked · 2 rankable rows
- Basis
- Provisional lane · 0 vs 2 weighted rows
- Reading
- Not comparable
Multilingual
Not comparable- Gemma 4 31B
- Not ranked
- GPT-5.2
- Not ranked
- Basis
- Provisional lane · 0 vs 0 weighted rows
- Reading
- Not comparable
| Category | Gemma 4 31B | GPT-5.2 | Basis | Reading |
|---|---|---|---|---|
| Agentic | 27.6Supported · #140/151 | 43.0Supported · #108/151 | Like-for-likeBenchAlign lane · 1 vs 4 public rows | GPT-5.2 leads · intervals overlap |
| Coding | 43.5Supported · #123/183 | 47.3Supported · #91/183 | Like-for-likeBenchAlign lane · 2 vs 3 public rows | GPT-5.2 leads · intervals overlap |
| Knowledge | 47.9Supported · #104/181 | 62.0Supported · #30/181 | Like-for-likeBenchAlign lane · 4 vs 1 public rows | GPT-5.2 leads · intervals overlap |
| Multimodal | 58.4#30/48 | 66.3#23/48 | Directional onlyProvisional lane · 1 vs 2 weighted rows | Directional only |
| Instruction following | 92.6#12/120 | 92.3#14/120 | Directional onlyProvisional lane · 0 vs 0 weighted rows | Directional only |
| Reasoning | 70.1Unranked · 2 rankable rows | 53.8Unranked · 3 rankable rows | Not comparableProvisional lane · 0 vs 1 weighted rows | Not comparable |
| Math | Not ranked | 57.5Unranked · 2 rankable rows | Not comparableProvisional lane · 0 vs 2 weighted rows | Not comparable |
| Multilingual | Not ranked | Not ranked | Not comparableProvisional lane · 0 vs 0 weighted rows | Not comparable |
Ranks count the models scored in each category’s lane, so the agentic, coding, and knowledge denominators (BenchAlign) differ from the provisional-lane categories. Unranked scores sit on the lane’s scale but never name a winner. Like-for-like rows within 0.5 points read as a practical tie.
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
MMMU-Pro
Multimodal
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-5.2
- $0.00875
- 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-5.2
- $0.1295
- 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-5.2
- $0.525
- Fits in one request
- Cached input priced at the published list-input rate
GPT-5.2 has no published cached-input rate, so cached tokens use its listed input rate. Gemma 4 31B 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 31BGPT-5.2
Context window
Maximum documented context; output-token limits may be lower.
Gemma 4 31B
GPT-5.2
400K
API model ID
Gemma 4 31B
gemma-4-31b-it
Google Gemma Gemini API guideGPT-5.2
Not sourced
Cached-input rate
A missing cached-input rate falls back to the listed input rate only in the stated workload estimate.
Gemma 4 31B
No comparable hosted API rate
GPT-5.2
Not published
Documented inputs
Gemma 4 31B
text, image
Google Gemma 4 model documentationGPT-5.2
Not sourced
Documented outputs
Gemma 4 31B
GPT-5.2
Not sourced
Provider availability
Gemma 4 31B
Generally Available · Gemini API, Google AI Studio, open weights
Google Gemma Gemini API guideGPT-5.2
Not sourced
Reasoning profile
Gemma 4 31B
Reasoning
GPT-5.2
Reasoning
Weight access
Gemma 4 31B
Open Weight
GPT-5.2
Proprietary
License
Gemma 4 31B
Open Weight
GPT-5.2
Proprietary
Release date
Gemma 4 31B
2026-04-02
GPT-5.2
2025-12-11
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-5.2 has the higher public score estimate, 65.45 versus 58.69, but the 90% score intervals overlap.
- Workload cost
- A complete comparable API-rate estimate is not available for both models.
- Context tradeoff
- GPT-5.2 has the larger documented window (400K).
Run the same representative tasks against both endpoints before changing production traffic.
Self-host vs API cost
Estimates at 50,000 req/day · 1000 tokens/req average.
Gemma 4 31B
API / mo$0
Self-host / mo$429
Break-even—
GPT-5.2
API / mo$11,813
Self-host / moNot listed
Break-even—
Proprietary model — self-hosting not applicable.
Benchmark evidence
The full public result ledger is available for audit without forcing a wide desktop table onto a phone.
Browse raw public benchmark evidence20 rows
Agentic
Gert Labs
Shared sourceGemma 4 31B35.26%GPT-5.246.54%GPT-5.2 leads this result
BrowseComp
Gemma 4 31B—GPT-5.265.8%SourceNot directly comparable
OSWorld-Verified
Gemma 4 31B—GPT-5.247.3%SourceNot directly comparable
JobBench
Gemma 4 31B—GPT-5.234.3%SourceNot directly comparable
Coding
SWE-Rebench
Gemma 4 31B41.6%SourceGPT-5.2—Not directly comparable
React Native Evals
Gemma 4 31B75.2%SourceGPT-5.2—Not directly comparable
SWE-bench Verified
Gemma 4 31B—GPT-5.280%SourceNot directly comparable
SWE-bench Pro
Gemma 4 31B—GPT-5.255.6%SourceNot directly comparable
Vibe Code Bench
Gemma 4 31B—GPT-5.253.50%SourceNot directly comparable
Reasoning
ARC-AGI-2
Gemma 4 31B—GPT-5.252.9%SourceNot directly comparable
Knowledge
Math
Multimodal
Frequently asked questions
Which is better, Gemma 4 31B or GPT-5.2?
GPT-5.2 has the higher public score estimate, 65.45 versus 58.69, 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 31B or GPT-5.2?
GPT-5.2 leads the public coding lane, 47.3 to 43.5, with Supported evidence for both models, although the 90% intervals overlap.
Which is better for agentic tasks, Gemma 4 31B or GPT-5.2?
GPT-5.2 leads the public agentic tasks lane, 43 to 27.6, with Supported evidence for both models, although the 90% intervals overlap.
Which costs less, Gemma 4 31B or GPT-5.2?
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-5.2?
GPT-5.2 has the larger documented context window: 400K, compared with 256K.
Related comparisons
Last updated September 4, 2026
Watch Gemma 4 31B vs GPT-5.2
One weekly email when material rank, price, or benchmark evidence changes make this matchup worth revisiting.
Read a sample issueJoin 2,000+ readers.