Enter opens the highlighted result. Esc closes.

Gemma 2 27B vs GPT-4

Updated October 21, 2026. 32 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Gemma 2 27B Replace GPT-4
Gemma 2 27BGoogle
Coding
No data
Overall
No data
Output $/M
$0.65
24
GPT-4OpenAI
Coding
No data
Overall
6.7
Output $/M
$60.00
8
vs boards won of 32 shared

Decision reading

Neither model has a coding score yet, so the two cannot be ranked on coding.

Gemma 2 27B has the edge where output tokens cost 92.3x less, and where the context window is larger, 8K against 8K tokens.

Model AGemma 2 27BGoogleNot benchmarkedNo coding benchmarkNo cost per solved taskModel BGPT-4OpenAINot benchmarkedNo coding benchmarkNo cost per solved task
Shared results
32
Gemma 2 27B only
4
GPT-4 only
13
Like-for-like categories
6 of 9

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

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

At a glance

Coding index
Gemma 2 27B not benchmarked, GPT-4 not benchmarked.
Widest category
Knowledge, 9.3 points apart on average over 5 shared benchmarks.
Widest single gap
DTBench: Gemma 2 27B 48.0% against GPT-4 62.7%, 14.7 points apart on a 0 to 100 scale.
Output price
Gemma 2 27B $0.65 against GPT-4 $60.00 per million output tokens, 92.3x apart.
Context window
Gemma 2 27B takes 8K tokens, GPT-4 8K.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
CategoryGemma 2 27BBarsGPT-4Ties
Coding20
Intelligence01
Reasoning12
Math02
Knowledge41
General172
All shared boards248
CodingLike-for-likeSix of these boards feed the Vibe Coding Index20
Coding benchmarks for Gemma 2 27B and GPT-4
BenchmarkGemma 2 27BGPT-4Leader
SWE-bench VerifiedNot measured22.4%Only GPT-4 measured
SWE-bench LiteNot measured39.7%Only GPT-4 measured
SWE-bench (full test set)Not measured12.5%Only GPT-4 measured
AA Coding IndexNot measured13.1Only GPT-4 measured
LMArena Text: coding12111209Gemma 2 27B
LMArena Text: software and IT12281195Gemma 2 27B
Neither model has a result on the other 32 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index01
Intelligence benchmarks for Gemma 2 27B and GPT-4
BenchmarkGemma 2 27BGPT-4Leader
AA Intelligence IndexNot measured6.7Only GPT-4 measured
Epoch Capabilities Index122.1125.7GPT-4
Agentic workNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Agentic work benchmarks for Gemma 2 27B and GPT-4
BenchmarkGemma 2 27BGPT-4Leader
Neither model has a result on 39 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index12
Reasoning benchmarks for Gemma 2 27B and GPT-4
BenchmarkGemma 2 27BGPT-4Leader
GPQA Diamond (Artificial Analysis run)Not measured34.9%Only GPT-4 measured
GPQA Diamond (Epoch AI run)36.5%Not measuredOnly Gemma 2 27B measured
BIG-Bench HardNot measured75.1%Only GPT-4 measured
LMArena Text: hard prompts11981200GPT-4
LMArena Text: expert11711149Gemma 2 27B
LMArena Text: hard prompts in English12061209GPT-4
Neither model has a result on the other 15 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for Gemma 2 27B and GPT-4
BenchmarkGemma 2 27BGPT-4Leader
OTIS Mock AIME1.4%Not measuredOnly Gemma 2 27B measured
MATH level 527.9%Not measuredOnly Gemma 2 27B measured
GSM8KNot measured92.0%Only GPT-4 measured
LMArena Text: math12121230GPT-4
LMArena Text: mathematical work12211228GPT-4
Neither model has a result on the other 8 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index41
Knowledge benchmarks for Gemma 2 27B and GPT-4
BenchmarkGemma 2 27BGPT-4Leader
LegalBench (Vals AI)68.0%Not measuredOnly Gemma 2 27B measured
MMLU Pro (Artificial Analysis run)Not measured56.2%Only GPT-4 measured
MMLU75.7%86.4%GPT-4
AI2 Reasoning ChallengeNot measured96.3%Only GPT-4 measured
LMArena Text: science12281177Gemma 2 27B
LMArena Text: medicine and healthcare11981145Gemma 2 27B
LMArena Text: legal and government12411191Gemma 2 27B
LMArena Text: business and finance12081145Gemma 2 27B
Neither model has a result on the other 9 boards here
Web designNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Web design benchmarks for Gemma 2 27B and GPT-4
BenchmarkGemma 2 27BGPT-4Leader
Neither model has a result on 9 boards here
GeneralLike-for-likePublic boards, shown as published, kept out of the index172
General benchmarks for Gemma 2 27B and GPT-4
BenchmarkGemma 2 27BGPT-4Leader
IFBenchNot measured33.2%Only GPT-4 measured
DTBench48.0%62.7%GPT-4
LMCA7.1%17.1%GPT-4
HellaSwagNot measured95.3%Only GPT-4 measured
WinoGrandeNot measured87.5%Only GPT-4 measured
LMArena Text12311206Gemma 2 27B
LMArena Text: instruction following12061201Gemma 2 27B
LMArena Text: creative writing12401191Gemma 2 27B
LMArena Text: English12421220Gemma 2 27B
LMArena Text: non-English12171194Gemma 2 27B
LMArena Text: Chinese12211184Gemma 2 27B
LMArena Text: German12091198Gemma 2 27B
LMArena Text: Japanese11751137Gemma 2 27B
LMArena Text: Korean11741087Gemma 2 27B
LMArena Text: French12481220Gemma 2 27B
LMArena Text: Spanish12281197Gemma 2 27B
LMArena Text: Russian12341195Gemma 2 27B
LMArena Text: Polish12771265Gemma 2 27B
LMArena Text: multi-turn12241206Gemma 2 27B
LMArena Text: longer queries12301190Gemma 2 27B
LMArena Text: writing and language12491219Gemma 2 27B
LMArena Text: entertainment and media12121184Gemma 2 27B
Neither model has a result on the other 24 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Gemma 2 27B and GPT-4
BenchmarkGemma 2 27BGPT-4Leader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

Gemma 2 27B$0.0010Fits in one request
GPT-4$0.060Fits in one request

Gemma 2 27B costs less

Repository review

50K input + 3K output tokens

Gemma 2 27B$0.034Exceeds the context window
GPT-4$1.68Exceeds the context window

Gemma 2 27B costs less

Agent loop

200K input + 10K output tokens

Gemma 2 27B$0.137Exceeds the context window
GPT-4$6.60Exceeds the context window

Gemma 2 27B costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of Gemma 2 27B and GPT-4
SpecificationGoogle Gemma 2 27BOpenAI GPT-4
MakerGoogleOpenAI
WeightsOpen weightsNo public weights listed
ReleasedJuly 13, 2024May 28, 2023
Context window8K8K
Input $/Mtok$0.65$30.00
Output $/Mtok$0.65$60.00
Coding indexNot benchmarkedNot benchmarked
Independent measurements behind it00
Cost per solved taskNot computableNot computable
SourceOpenRouter listingOpenRouter listing

Questions

Is Gemma 2 27B better than GPT-4 for coding?

Neither model has a coding score yet, so the two cannot be ranked on coding. Gemma 2 27B has the edge where output tokens cost 92.3x less, and where the context window is larger, 8K against 8K tokens.

Which is cheaper, Gemma 2 27B or GPT-4?

Gemma 2 27B costs $0.65 per million input tokens and $0.65 per million output tokens; GPT-4 costs $30.00 and $60.00. Cost per solved task is known for neither of them.

Which has the bigger context window, Gemma 2 27B or GPT-4?

Gemma 2 27B, with 8K tokens against 8K.

How many benchmarks is the Gemma 2 27B vs GPT-4 comparison based on?

32 benchmarks have a result for both models as of October 21, 2026. Gemma 2 27B's coding index rests on 0 independent benchmarks and GPT-4's on 0. Where only one model has a result, the page says so instead of naming a winner.

Compare next