LLMBenchmark.ai

Enter opens the highlighted result. Esc closes.

Gemini 2.5 Flash vs Qwen3.8 27B

Updated October 21, 2026. 48 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.

Replace Gemini 2.5 Flash Replace Qwen3.8 27B
Gemini 2.5 FlashGoogle
Coding
28.7
Overall
15.5
Output $/M
$2.50
7
Qwen3.8 27BQwen
Coding
73.6
Overall
33.7
Output $/M
$1.35
41
vs boards won of 48 shared

Decision reading

Pick Qwen3.8 27B for coding: it scores 73.6 on the Vibe Coding Index against 28.7 for Gemini 2.5 Flash.

Pick Gemini 2.5 Flash when a solved task costs $0.466 against $1.26, or when the context window is larger, 1.05M against 1M tokens.

Model AGemini 2.5 FlashGoogle28.7/1001 independent benchmark$0.466 per solved taskModel BQwen3.8 27BQwen73.6/1002 independent benchmarks$1.26 per solved task
Shared results
48
Gemini 2.5 Flash only
21
Qwen3.8 27B only
37
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
Gemini 2.5 Flash 28.7 from 1 benchmark, Qwen3.8 27B 73.6 from 2 benchmarks.
Widest category
Reasoning, 18.2 points apart on average over 9 shared benchmarks.
Widest single gap
ARC-AGI: Gemini 2.5 Flash 33.3% against Qwen3.8 27B 87.5%, 54.2 points apart on a 0 to 100 scale.
Output price
Gemini 2.5 Flash $2.50 against Qwen3.8 27B $1.35 per million output tokens, 1.9x apart.
Context window
Gemini 2.5 Flash takes 1.05M tokens, Qwen3.8 27B 1M.

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
CategoryGemini 2.5 FlashBarsQwen3.8 27BTies
Coding02
Intelligence02
Reasoning09
Math02
Knowledge14
General622
All shared boards741
CodingLike-for-likeFour of these boards feed the Vibe Coding Index02
Coding benchmarks for Gemini 2.5 Flash and Qwen3.8 27B
BenchmarkGemini 2.5 FlashQwen3.8 27BLeader
SWE-bench Verified28.7%Not measuredOnly Gemini 2.5 Flash measured
LiveBench codingNot measured86.8%Only Qwen3.8 27B measured
LMArena WebDevNot measured1593Only Qwen3.8 27B measured
AA Coding IndexNot measured68.1Only Qwen3.8 27B measured
LiveCodeBench (Vals AI)Not measured84.0%Only Qwen3.8 27B measured
SWE-bench Verified (Vals AI)Not measured86.0%Only Qwen3.8 27B measured
Vibe Code Bench (Vals AI)Not measured64.8%Only Qwen3.8 27B measured
Code migration (Vals AI)Not measured14.2%Only Qwen3.8 27B measured
ProgramBench (Vals AI)Not measured0.0%Only Qwen3.8 27B measured
IOI (Vals AI)Not measured39.1%Only Qwen3.8 27B measured
Terminal-Bench 2.1 (Vals AI)Not measured58.4%Only Qwen3.8 27B measured
SkillsBench (Vals AI)Not measured38.1%Only Qwen3.8 27B measured
CyberBench (Vals AI)Not measured67.8%Only Qwen3.8 27B measured
LiveCodeBench (Artificial Analysis run)69.5%Not measuredOnly Gemini 2.5 Flash measured
SciCodeNot measured46.6%Only Qwen3.8 27B measured
Terminal-Bench Hard (Artificial Analysis run)13.6%Not measuredOnly Gemini 2.5 Flash measured
Terminal-Bench17.1%Not measuredOnly Gemini 2.5 Flash measured
LMArena Text: coding14231483Qwen3.8 27B
LMArena Text: software and IT14251478Qwen3.8 27B
LMArena WebDev: ReactNot measured1602Only Qwen3.8 27B measured
LMArena WebDev: HTMLNot measured1570Only Qwen3.8 27B measured
LMArena WebDev: from an imageNot measured1577Only Qwen3.8 27B measured
Neither model has a result on the other 16 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index02
Intelligence benchmarks for Gemini 2.5 Flash and Qwen3.8 27B
BenchmarkGemini 2.5 FlashQwen3.8 27BLeader
AA Intelligence Index15.533.7Qwen3.8 27B
Epoch Capabilities Index142.9151.7Qwen3.8 27B
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index13 boards, none shared
Agentic work benchmarks for Gemini 2.5 Flash and Qwen3.8 27B
BenchmarkGemini 2.5 FlashQwen3.8 27BLeader
AA Agentic IndexNot measured45.8Only Qwen3.8 27B measured
LiveBench researchNot measured59.4%Only Qwen3.8 27B measured
LiveBench browserNot measured32.0%Only Qwen3.8 27B measured
LiveBench long agenticNot measured27.9%Only Qwen3.8 27B measured
Finance Agent (Vals AI)Not measured48.5%Only Qwen3.8 27B measured
Tax Agent Bench (Vals AI)Not measured30.9%Only Qwen3.8 27B measured
Legal Research Bench (Vals AI)Not measured36.1%Only Qwen3.8 27B measured
Excel Modeling Benchmark (Vals AI)Not measured59.7%Only Qwen3.8 27B measured
Terminal-Bench-Science (Vals AI)Not measured0.0%Only Qwen3.8 27B measured
τ²-Bench Telecom31.6%Not measuredOnly Gemini 2.5 Flash measured
APEX-AgentsNot measured47.5%Only Qwen3.8 27B measured
BALROG33.5%Not measuredOnly Gemini 2.5 Flash measured
Vending-Bench 2$549Not measuredOnly Gemini 2.5 Flash measured
Neither model has a result on the other 26 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index09
Reasoning benchmarks for Gemini 2.5 Flash and Qwen3.8 27B
BenchmarkGemini 2.5 FlashQwen3.8 27BLeader
GPQA DiamondNot measured88.9%Only Qwen3.8 27B measured
LiveBench reasoningNot measured80.0%Only Qwen3.8 27B measured
MMMU Pro (Vals AI)Not measured83.9%Only Qwen3.8 27B measured
GPQA Diamond (Artificial Analysis run)79.0%90.5%Qwen3.8 27B
Humanity's Last Exam (Artificial Analysis run)12.1%33.9%Qwen3.8 27B
ARC-AGI33.3%87.5%Qwen3.8 27B
ARC-AGI-22.5%42.4%Qwen3.8 27B
SimpleBench41.2%Not measuredOnly Gemini 2.5 Flash measured
LMArena Text: hard prompts14221459Qwen3.8 27B
LMArena Text: expert14261479Qwen3.8 27B
LMArena Text: hard prompts in English14201461Qwen3.8 27B
LMArena Vision: diagrams12451285Qwen3.8 27B
LMArena Vision: homework12441295Qwen3.8 27B
Neither model has a result on the other 8 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index02
Math benchmarks for Gemini 2.5 Flash and Qwen3.8 27B
BenchmarkGemini 2.5 FlashQwen3.8 27BLeader
AA Math Index73.3Not measuredOnly Gemini 2.5 Flash measured
AIME 2025 (Artificial Analysis run)73.3%Not measuredOnly Gemini 2.5 Flash measured
ProofBenchNot measured16.0%Only Qwen3.8 27B measured
LMArena Text: math14111455Qwen3.8 27B
LMArena Text: mathematical work14221458Qwen3.8 27B
Neither model has a result on the other 8 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index14
Knowledge benchmarks for Gemini 2.5 Flash and Qwen3.8 27B
BenchmarkGemini 2.5 FlashQwen3.8 27BLeader
MMLU ProNot measured84.3%Only Qwen3.8 27B measured
LegalBench (Vals AI)Not measured82.4%Only Qwen3.8 27B measured
TaxEval (Vals AI)Not measured70.8%Only Qwen3.8 27B measured
MortgageTax (Vals AI)Not measured64.9%Only Qwen3.8 27B measured
MedCode (Vals AI)40.4%28.7%Gemini 2.5 Flash
MMLU Pro (Artificial Analysis run)83.2%Not measuredOnly Gemini 2.5 Flash measured
LMArena Text: science14331467Qwen3.8 27B
LMArena Text: medicine and healthcare14281437Qwen3.8 27B
LMArena Text: legal and government14301447Qwen3.8 27B
LMArena Text: business and finance14041438Qwen3.8 27B
Neither model has a result on the other 7 boards here
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
Web design benchmarks for Gemini 2.5 Flash and Qwen3.8 27B
BenchmarkGemini 2.5 FlashQwen3.8 27BLeader
Design Arena: websites1074Not measuredOnly Gemini 2.5 Flash measured
Design Arena: UI components1005Not measuredOnly Gemini 2.5 Flash measured
Design Arena: game dev1005Not measuredOnly Gemini 2.5 Flash measured
Design Arena: data viz1060Not measuredOnly Gemini 2.5 Flash measured
Design Arena: 3D1021Not measuredOnly Gemini 2.5 Flash measured
Design Arena: all code categories1059Not measuredOnly Gemini 2.5 Flash measured
Design Arena: SVG1018Not measuredOnly Gemini 2.5 Flash measured
LiveBench designNot measured75.9%Only Qwen3.8 27B measured
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index622
General benchmarks for Gemini 2.5 Flash and Qwen3.8 27B
BenchmarkGemini 2.5 FlashQwen3.8 27BLeader
LiveBench data analysisNot measured76.6%Only Qwen3.8 27B measured
SAGE (Vals AI)44.8%52.4%Qwen3.8 27B
MedScribe (Vals AI)83.0%83.8%Qwen3.8 27B
IFBench50.3%Not measuredOnly Gemini 2.5 Flash measured
Long context reasoning (AA-LCR)65.3%82.0%Qwen3.8 27B
DTBench76.5%88.0%Qwen3.8 27B
LMCA27.5%41.4%Qwen3.8 27B
Surface Evolver BenchNot measured45.0%Only Qwen3.8 27B measured
LMArena Text14171440Qwen3.8 27B
LMArena Text: instruction following14051436Qwen3.8 27B
LMArena Text: creative writing14001383Gemini 2.5 Flash
LMArena Vision12351269Qwen3.8 27B
LMArena Text: English14181448Qwen3.8 27B
LMArena Text: non-English14091428Qwen3.8 27B
LMArena Text: Chinese14491501Qwen3.8 27B
LMArena Text: German14171443Qwen3.8 27B
LMArena Text: Japanese13961387Gemini 2.5 Flash
LMArena Text: Korean13851384Gemini 2.5 Flash
LMArena Text: French14321465Qwen3.8 27B
LMArena Text: Spanish14221449Qwen3.8 27B
LMArena Text: Russian14151415Qwen3.8 27B
LMArena Text: Polish14131439Qwen3.8 27B
LMArena Text: multi-turn14081440Qwen3.8 27B
LMArena Text: longer queries14191448Qwen3.8 27B
LMArena Text: writing and language14051405Gemini 2.5 Flash
LMArena Text: entertainment and media13931393Gemini 2.5 Flash
LMArena Vision: English12391267Qwen3.8 27B
LMArena Vision: Chinese12841326Qwen3.8 27B
LMArena Vision: reading text12371281Qwen3.8 27B
LMArena Vision: humor12251235Qwen3.8 27B
LMArena Vision: captioning1277Not measuredOnly Gemini 2.5 Flash measured
LMArena Vision: recognizing things1252Not measuredOnly Gemini 2.5 Flash measured
LMArena Vision: creative writing12411237Gemini 2.5 Flash
Neither model has a result on the other 13 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for Gemini 2.5 Flash and Qwen3.8 27B
BenchmarkGemini 2.5 FlashQwen3.8 27BLeader
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

Gemini 2.5 Flash$0.0015Fits in one request
Qwen3.8 27B$0.0007Fits in one request

Qwen3.8 27B costs less

Repository review

50K input + 3K output tokens

Gemini 2.5 Flash$0.022Fits in one request
Qwen3.8 27B$0.0060Fits in one request

Qwen3.8 27B costs less

Agent loop

200K input + 10K output tokens

Gemini 2.5 Flash$0.085Fits in one request
Qwen3.8 27B$0.021Fits in one request

Qwen3.8 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 Gemini 2.5 Flash and Qwen3.8 27B
SpecificationGoogle Gemini 2.5 FlashQwen3.8 27B
MakerGoogleQwen
WeightsNo public weights listedOpen weights
ReleasedJune 17, 2025August 14, 2026
Context window1.05M1M
Input $/Mtok$0.30$0.04
Output $/Mtok$2.50$1.35
Coding index28.773.6
Independent measurements behind it12
Cost per solved task$0.466$1.26
SourceOpenRouter listingOpenRouter listing

Questions

Is Gemini 2.5 Flash better than Qwen3.8 27B for coding?

Pick Qwen3.8 27B for coding: it scores 73.6 on the Vibe Coding Index against 28.7 for Gemini 2.5 Flash. Pick Gemini 2.5 Flash when a solved task costs $0.466 against $1.26, or when the context window is larger, 1.05M against 1M tokens.

Which is cheaper, Gemini 2.5 Flash or Qwen3.8 27B?

Gemini 2.5 Flash costs $0.30 per million input tokens and $2.50 per million output tokens; Qwen3.8 27B costs $0.04 and $1.35. Per solved SWE-bench Verified task, Gemini 2.5 Flash costs $0.466 and Qwen3.8 27B $1.26.

Which has the bigger context window, Gemini 2.5 Flash or Qwen3.8 27B?

Gemini 2.5 Flash, with 1.05M tokens against 1M.

How many benchmarks is the Gemini 2.5 Flash vs Qwen3.8 27B comparison based on?

48 benchmarks have a result for both models as of October 21, 2026. Gemini 2.5 Flash's coding index rests on 1 independent benchmark and Qwen3.8 27B's on 2. Where only one model has a result, the page says so instead of naming a winner.

Compare next