Gemini 3 Flash Preview vs Gemini 4 Carbon
Gemini 4 Carbon leads Gemini 3 Flash Preview on coding, 86.6 to 43.7 on the Vibe Coding Index, and Gemini 3 Flash Preview costs 3.3x less per output token ($3.00 vs $10.00 per million). Updated November 14, 2026; 22 benchmarks have a result for both.
vs boards won of 22 shared
Gemini 3 Flash Preview or Gemini 4 Carbon: which to pick
Pick Gemini 4 Carbon for coding: it scores 86.6 on the Vibe Coding Index against 43.7 for Gemini 3 Flash Preview.
Pick Gemini 3 Flash Preview for output tokens 3.3x cheaper ($3.00 vs $10.00 per million).
Model AGemini 3 Flash PreviewGoogle43.7/1002 independent benchmarks$0.37 per solved taskModel BGemini 4 CarbonGoogle86.6/1004 independent benchmarksNo cost per solved task
- Shared results
- 22
- Gemini 3 Flash Preview only
- 47
- Gemini 4 Carbon only
- 51
- Like-for-like categories
- 7 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
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.
Coding
Vibe Coding Index, six public coding benchmarks
Gemini 4 Carbon
Gemini 4 Carbon scores 86.6 against 43.7.
2 and 4 benchmarks behind the scores
Intelligence
AA Intelligence Index
No call
Only Gemini 4 Carbon has a AA Intelligence Index result, 60.0.
Directional only
Reasoning
GPQA Diamond
No call
Only Gemini 3 Flash Preview has a GPQA Diamond result, 87.9%.
Directional only
Math
AIME
No call
Only Gemini 3 Flash Preview has a AIME result, 95.6%.
Directional only
Knowledge
MMLU Pro
No call
Only Gemini 3 Flash Preview has a MMLU Pro result, 88.6%.
Directional only
Web design
Design Arena: websites
Gemini 4 Carbon
Gemini 4 Carbon leads, 1319 against 1188.
Same board, both measured
General
Vals Index
No call
Only Gemini 4 Carbon has a Vals Index result, 70.5%.
Directional only
Long documents
Listed context window
No call
Both take 1.05M tokens.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Gemini 3 Flash Preview
Gemini 3 Flash Preview $0.0020, Gemini 4 Carbon $0.0070 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Gemini 3 Flash Preview has the number, $0.37.
Directional only
At a glance
- Coding index
- Gemini 3 Flash Preview 43.7 from 2 benchmarks, Gemini 4 Carbon 86.6 from 4 benchmarks.
- Widest category
- Coding, 51.2 points apart on average over 2 shared benchmarks.
- Widest single gap
- FrontierMath, tier 4: Gemini 3 Flash Preview 17.1% against Gemini 4 Carbon 95.1%, 78.0 points apart on a 0 to 100 scale.
- Output price
- Gemini 3 Flash Preview $3.00 against Gemini 4 Carbon $10.00 per million output tokens, 3.3x apart.
- Context window
- Both take 1.05M tokens.
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.
| Category | Gemini 3 Flash Preview | Bars | Gemini 4 Carbon | Ties |
|---|---|---|---|---|
| Coding | 0 | 2 | ||
| Agentic work | 0 | 4 | ||
| Reasoning | 0 | 6 | ||
| Math | 0 | 3 | ||
| Knowledge | 0 | 2 | ||
| Web design | 0 | 4 | ||
| General | 0 | 1 | ||
| All shared boards | 0 | 22 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index02
| Benchmark | Gemini 3 Flash Preview | Gemini 4 Carbon | Leader |
|---|---|---|---|
| SWE-bench Verified | 75.8% | Not measured | Only Gemini 3 Flash Preview measured |
| LiveBench coding | Not measured | 91.0% | Only Gemini 4 Carbon measured |
| LMArena WebDev | Not measured | 1768 | Only Gemini 4 Carbon measured |
| LiveCodeBench (Vals AI) | 85.6% | Not measured | Only Gemini 3 Flash Preview measured |
| SWE-bench Verified (Vals AI) | 75.0% | Not measured | Only Gemini 3 Flash Preview measured |
| Terminal-Bench 4 (Vals AI) | Not measured | 63.1% | Only Gemini 4 Carbon measured |
| Vibe Code Bench (Vals AI) | 20.2% | 89.1% | Gemini 4 Carbon |
| Code migration (Vals AI) | 6.4% | Not measured | Only Gemini 3 Flash Preview measured |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Gemini 3 Flash Preview measured |
| Terminal-Bench 2.1 (Vals AI) | 53.9% | Not measured | Only Gemini 3 Flash Preview measured |
| LiveCodeBench (Artificial Analysis run) | 90.8% | Not measured | Only Gemini 3 Flash Preview measured |
| Terminal-Bench Hard (Artificial Analysis run) | 38.6% | Not measured | Only Gemini 3 Flash Preview measured |
| Terminal-Bench 4.0 (Artificial Analysis run) | Not measured | 65.2% | Only Gemini 4 Carbon measured |
| FrontierCode 2.0 | Not measured | 40.6% | Only Gemini 4 Carbon measured |
| Terminal-Bench | 64.3% | Not measured | Only Gemini 3 Flash Preview measured |
| FrontierSWE | Not measured | 61.2% | Only Gemini 4 Carbon measured |
| GSO-Bench | 9.8% | Not measured | Only Gemini 3 Flash Preview measured |
| WeirdML | 61.6% | Not measured | Only Gemini 3 Flash Preview measured |
| ALE-Bench | 1367.2 | 2374.7 | Gemini 4 Carbon |
| LMArena Text: coding | Not measured | 1564 | Only Gemini 4 Carbon measured |
| LMArena Text: software and IT | Not measured | 1552 | Only Gemini 4 Carbon measured |
| LMArena WebDev: React | Not measured | 1792 | Only Gemini 4 Carbon measured |
| LMArena WebDev: HTML | Not measured | 1770 | Only Gemini 4 Carbon measured |
| SciCode | 12.7% | Not measured | Only Gemini 3 Flash Preview measured |
| CursorBench 4.0 | Not measured | 60.4% | Only Gemini 4 Carbon measured |
| Neither model has a result on the other 27 boards here | |||
IntelligenceDirectional onlyPublic boards, shown as published, kept out of the index2 boards, none shared
| Benchmark | Gemini 3 Flash Preview | Gemini 4 Carbon | Leader |
|---|---|---|---|
| AA Intelligence Index | Not measured | 60.0 | Only Gemini 4 Carbon measured |
| Epoch Capabilities Index | Not measured | 169.8 | Only Gemini 4 Carbon measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index04
| Benchmark | Gemini 3 Flash Preview | Gemini 4 Carbon | Leader |
|---|---|---|---|
| Design Arena: agents building web apps | 1087 | Not measured | Only Gemini 3 Flash Preview measured |
| Design Arena: agents building full-stack apps | 1037 | Not measured | Only Gemini 3 Flash Preview measured |
| Design Arena: agents building mobile apps | 1082 | 1284 | Gemini 4 Carbon |
| Design Arena: agents building Android apps | 1034 | Not measured | Only Gemini 3 Flash Preview measured |
| Design Arena: agents building Godot games | 1161 | Not measured | Only Gemini 3 Flash Preview measured |
| Design Arena: PowerPoint slides | 1036 | Not measured | Only Gemini 3 Flash Preview measured |
| Design Arena: spreadsheets | Not measured | 1224 | Only Gemini 4 Carbon measured |
| LiveBench research | Not measured | 92.5% | Only Gemini 4 Carbon measured |
| LiveBench browser | Not measured | 80.0% | Only Gemini 4 Carbon measured |
| LiveBench long agentic | Not measured | 72.7% | Only Gemini 4 Carbon measured |
| Finance Agent (Vals AI) | 42.5% | Not measured | Only Gemini 3 Flash Preview measured |
| Tax Agent Bench (Vals AI) | 11.8% | 51.8% | Gemini 4 Carbon |
| Legal Research Bench (Vals AI) | 18.3% | 55.8% | Gemini 4 Carbon |
| Harvey's Legal Agent Benchmark (Vals AI) | Not measured | 17.5% | Only Gemini 4 Carbon measured |
| Excel Modeling Benchmark (Vals AI) | 26.0% | Not measured | Only Gemini 3 Flash Preview measured |
| τ²-Bench Telecom | 80.4% | Not measured | Only Gemini 3 Flash Preview measured |
| APEX-Agents | Not measured | 84.2% | Only Gemini 4 Carbon measured |
| BALROG | 48.1% | Not measured | Only Gemini 3 Flash Preview measured |
| DeepResearch Bench | 49.8% | Not measured | Only Gemini 3 Flash Preview measured |
| Vending-Bench 2 | $3,635 | $16,484 | Gemini 4 Carbon |
| MCP Atlas | 62.0% | Not measured | Only Gemini 3 Flash Preview measured |
| Neither model has a result on the other 23 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index06
| Benchmark | Gemini 3 Flash Preview | Gemini 4 Carbon | Leader |
|---|---|---|---|
| GPQA Diamond | 87.9% | Not measured | Only Gemini 3 Flash Preview measured |
| LiveBench reasoning | Not measured | 90.6% | Only Gemini 4 Carbon measured |
| MMMU Pro (Vals AI) | 87.6% | Not measured | Only Gemini 3 Flash Preview measured |
| GPQA Diamond (Artificial Analysis run) | 89.8% | Not measured | Only Gemini 3 Flash Preview measured |
| Humanity's Last Exam (Artificial Analysis run) | 36.6% | Not measured | Only Gemini 3 Flash Preview measured |
| HLE-Diamond (Artificial Analysis run) | Not measured | 69.1% | Only Gemini 4 Carbon measured |
| CritPt v2, hard subset (Artificial Analysis run) | Not measured | 69.0% | Only Gemini 4 Carbon measured |
| GPQA Diamond (Epoch AI run) | 89.4% | 96.5% | Gemini 4 Carbon |
| ARC-AGI | 84.7% | 98.5% | Gemini 4 Carbon |
| ARC-AGI-2 | 33.6% | 94.2% | Gemini 4 Carbon |
| SimpleBench | 61.1% | 92.6% | Gemini 4 Carbon |
| Chess puzzles | 40.0% | 78.0% | Gemini 4 Carbon |
| Mystery game puzzles | 26.0% | 73.0% | Gemini 4 Carbon |
| EBR-bench | Not measured | 87.6% | Only Gemini 4 Carbon measured |
| VPCT | 72.6% | Not measured | Only Gemini 3 Flash Preview measured |
| LMArena Text: hard prompts | Not measured | 1551 | Only Gemini 4 Carbon measured |
| LMArena Text: expert | Not measured | 1582 | Only Gemini 4 Carbon measured |
| LMArena Text: hard prompts in English | Not measured | 1545 | Only Gemini 4 Carbon measured |
| Neither model has a result on the other 8 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index03
| Benchmark | Gemini 3 Flash Preview | Gemini 4 Carbon | Leader |
|---|---|---|---|
| AIME | 95.6% | Not measured | Only Gemini 3 Flash Preview measured |
| MGSM (Vals AI) | 93.3% | Not measured | Only Gemini 3 Flash Preview measured |
| AIME 2025 (Artificial Analysis run) | 97.0% | Not measured | Only Gemini 3 Flash Preview measured |
| OTIS Mock AIME | 95.6% | 100.0% | Gemini 4 Carbon |
| FrontierMath, tiers 1 to 3 | 51.2% | 90.5% | Gemini 4 Carbon |
| FrontierMath, tier 4 | 17.1% | 95.1% | Gemini 4 Carbon |
| ProofBench | 15.0% | Not measured | Only Gemini 3 Flash Preview measured |
| LMArena Text: mathematical work | Not measured | 1538 | Only Gemini 4 Carbon measured |
| MathArena | Not measured | 85.4% | Only Gemini 4 Carbon measured |
| Neither model has a result on the other 5 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index02
| Benchmark | Gemini 3 Flash Preview | Gemini 4 Carbon | Leader |
|---|---|---|---|
| MMLU Pro | 88.6% | Not measured | Only Gemini 3 Flash Preview measured |
| LegalBench (Vals AI) | 86.9% | Not measured | Only Gemini 3 Flash Preview measured |
| CaseLaw (Vals AI) | 55.8% | Not measured | Only Gemini 3 Flash Preview measured |
| CorpFin (Vals AI) | 66.4% | Not measured | Only Gemini 3 Flash Preview measured |
| TaxEval (Vals AI) | 73.9% | Not measured | Only Gemini 3 Flash Preview measured |
| MortgageTax (Vals AI) | 68.7% | Not measured | Only Gemini 3 Flash Preview measured |
| MedQA (Vals AI) | 95.8% | Not measured | Only Gemini 3 Flash Preview measured |
| MedCode (Vals AI) | 55.9% | Not measured | Only Gemini 3 Flash Preview measured |
| MMLU Pro (Artificial Analysis run) | 89.0% | Not measured | Only Gemini 3 Flash Preview measured |
| AA-Omniscience | Not measured | 53.5 | Only Gemini 4 Carbon measured |
| SimpleQA Verified | 66.8% | 77.7% | Gemini 4 Carbon |
| GeoBench | 88.0% | Not measured | Only Gemini 3 Flash Preview measured |
| LMArena Text: science | Not measured | 1535 | Only Gemini 4 Carbon measured |
| LMArena Text: medicine and healthcare | Not measured | 1525 | Only Gemini 4 Carbon measured |
| LMArena Text: legal and government | Not measured | 1542 | Only Gemini 4 Carbon measured |
| LMArena Text: business and finance | Not measured | 1520 | Only Gemini 4 Carbon measured |
| FACTS Benchmark Suite | 63.8% | 78.4% | Gemini 4 Carbon |
| PRBench Legal | Not measured | 53.3% | Only Gemini 4 Carbon measured |
| Neither model has a result on the other 3 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index04
| Benchmark | Gemini 3 Flash Preview | Gemini 4 Carbon | Leader |
|---|---|---|---|
| Design Arena: websites | 1188 | 1319 | Gemini 4 Carbon |
| Design Arena: UI components | Not measured | 1357 | Only Gemini 4 Carbon measured |
| Design Arena: game dev | 1155 | 1326 | Gemini 4 Carbon |
| Design Arena: 3D | 1163 | 1434 | Gemini 4 Carbon |
| Design Arena: all code categories | 1179 | 1348 | Gemini 4 Carbon |
| LiveBench design | Not measured | 80.2% | Only Gemini 4 Carbon measured |
| Neither model has a result on the other 3 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index01
| Benchmark | Gemini 3 Flash Preview | Gemini 4 Carbon | Leader |
|---|---|---|---|
| Vals Index | Not measured | 70.5% | Only Gemini 4 Carbon measured |
| LiveBench data analysis | Not measured | 86.5% | Only Gemini 4 Carbon measured |
| MedScribe (Vals AI) | 69.9% | Not measured | Only Gemini 3 Flash Preview measured |
| Vals Multimodal Index | 52.2% | Not measured | Only Gemini 3 Flash Preview measured |
| IFBench | 78.0% | Not measured | Only Gemini 3 Flash Preview measured |
| Long context reasoning (AA-LCR) | 78.0% | 79.3% | Gemini 4 Carbon |
| GDPval-AA | Not measured | 1733 | Only Gemini 4 Carbon measured |
| DTBench | 89.1% | Not measured | Only Gemini 3 Flash Preview measured |
| LMCA | 43.1% | Not measured | Only Gemini 3 Flash Preview measured |
| Furniture assembly | Not measured | 77.5% | Only Gemini 4 Carbon measured |
| LMArena Text | Not measured | 1538 | Only Gemini 4 Carbon measured |
| LMArena Text: instruction following | Not measured | 1539 | Only Gemini 4 Carbon measured |
| LMArena Text: creative writing | Not measured | 1533 | Only Gemini 4 Carbon measured |
| LMArena Text: English | Not measured | 1530 | Only Gemini 4 Carbon measured |
| LMArena Text: non-English | Not measured | 1534 | Only Gemini 4 Carbon measured |
| LMArena Text: Chinese | Not measured | 1613 | Only Gemini 4 Carbon measured |
| LMArena Text: Russian | Not measured | 1560 | Only Gemini 4 Carbon measured |
| LMArena Text: multi-turn | Not measured | 1569 | Only Gemini 4 Carbon measured |
| LMArena Text: longer queries | Not measured | 1549 | Only Gemini 4 Carbon measured |
| LMArena Text: writing and language | Not measured | 1542 | Only Gemini 4 Carbon measured |
| LMArena Text: entertainment and media | Not measured | 1527 | Only Gemini 4 Carbon measured |
| Neither model has a result on the other 28 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Gemini 3 Flash Preview | Gemini 4 Carbon | Leader |
|---|---|---|---|
| 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 3 Flash Preview$0.0020Fits in one request
Gemini 4 Carbon$0.0070Fits in one request
Gemini 3 Flash Preview costs less
Repository review
50K input + 3K output tokens
Gemini 3 Flash Preview$0.034Fits in one request
Gemini 4 Carbon$0.130Fits in one request
Gemini 3 Flash Preview costs less
Agent loop
200K input + 10K output tokens
Gemini 3 Flash Preview$0.130Fits in one request
Gemini 4 Carbon$0.500Fits in one request
Gemini 3 Flash Preview 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.
| Specification | Google Gemini 3 Flash Preview | Google Gemini 4 Carbon |
|---|---|---|
| Maker | ||
| Weights | No public weights listed | No public weights listed |
| Released | December 17, 2025 | November 10, 2026 |
| Context window | 1.05M | 1.05M |
| Input $/Mtok | $0.50 | $2.00 |
| Output $/Mtok | $3.00 | $10.00 |
| Coding index | 43.7 | 86.6 |
| Independent measurements behind it | 2 | 4 |
| Cost per solved task | $0.37 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Gemini 3 Flash Preview better than Gemini 4 Carbon for coding?
Pick Gemini 4 Carbon for coding: it scores 86.6 on the Vibe Coding Index against 43.7 for Gemini 3 Flash Preview. Pick Gemini 3 Flash Preview for output tokens 3.3x cheaper ($3.00 vs $10.00 per million).
Which is cheaper, Gemini 3 Flash Preview or Gemini 4 Carbon?
Gemini 3 Flash Preview costs $0.50 per million input tokens and $3.00 per million output tokens; Gemini 4 Carbon costs $2.00 and $10.00. Cost per solved task is known for Gemini 3 Flash Preview only.
Which has the bigger context window, Gemini 3 Flash Preview or Gemini 4 Carbon?
Neither: both take 1.05M tokens.
How many benchmarks is the Gemini 3 Flash Preview vs Gemini 4 Carbon comparison based on?
22 benchmarks have a result for both models as of November 14, 2026. Gemini 3 Flash Preview's coding index rests on 2 independent benchmarks and Gemini 4 Carbon's on 4. Where only one model has a result, the page says so instead of naming a winner.