Gemini 3.5 Flash vs Phi 4
Updated October 21, 2026. 37 benchmarks have a result for both. Coding is ranked on the Vibe Coding Index; every other board is shown as published.
vs boards won of 37 shared
Decision reading
Only Gemini 3.5 Flash has a coding score, 39.8 on the Vibe Coding Index; Phi 4 has none, so the two cannot be ranked on coding.
Phi 4 has the edge where output tokens cost 64.3x less.
Model AGemini 3.5 FlashGoogle39.8/1004 independent benchmarks$1.21 per solved taskModel BPhi 4MicrosoftNot benchmarkedNo coding benchmarkNo cost per solved task
- Shared results
- 37
- Gemini 3.5 Flash only
- 73
- Phi 4 only
- 8
- Like-for-like categories
- 7 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.
Coding
Vibe Coding Index, six public coding benchmarks
No call
Only Gemini 3.5 Flash has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
Gemini 3.5 Flash
Gemini 3.5 Flash leads, 33.6 against 5.9.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only Gemini 3.5 Flash has a GPQA Diamond result, 92.7%.
Directional only
Knowledge
MMLU Pro
No call
Only Gemini 3.5 Flash has a MMLU Pro result, 89.5%.
Directional only
Web design
Design Arena: websites
No call
Only Gemini 3.5 Flash has a Design Arena: websites result, 1268.
Directional only
Long documents
Listed context window
Gemini 3.5 Flash
Gemini 3.5 Flash takes 1.05M tokens against 16K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Phi 4
Gemini 3.5 Flash $0.0060, Phi 4 $0.0001 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Gemini 3.5 Flash has the number, $1.21.
Directional only
At a glance
- Coding index
- Gemini 3.5 Flash 39.8 from 4 benchmarks, Phi 4 not benchmarked.
- Widest category
- Agentic work, 95.6 points apart on average over 1 shared benchmark.
- Widest single gap
- τ²-Bench Telecom: Gemini 3.5 Flash 95.6% against Phi 4 0.0%, 95.6 points apart on a 0 to 100 scale.
- Output price
- Gemini 3.5 Flash $9.00 against Phi 4 $0.14 per million output tokens, 64.3x apart.
- Context window
- Gemini 3.5 Flash takes 1.05M tokens, Phi 4 16K.
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 3.5 Flash | Bars | Phi 4 | Ties |
|---|---|---|---|---|
| Coding | 3 | 0 | ||
| Intelligence | 2 | 0 | ||
| Agentic work | 1 | 0 | ||
| Reasoning | 6 | 0 | ||
| Math | 3 | 0 | ||
| Knowledge | 4 | 0 | ||
| General | 18 | 0 | ||
| All shared boards | 37 | 0 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index30
| Benchmark | Gemini 3.5 Flash | Phi 4 | Leader |
|---|---|---|---|
| LMArena WebDev | 1499 | Not measured | Only Gemini 3.5 Flash measured |
| LiveCodeBench (Vals AI) | 87.6% | Not measured | Only Gemini 3.5 Flash measured |
| SWE-bench Verified (Vals AI) | 78.8% | Not measured | Only Gemini 3.5 Flash measured |
| Terminal-Bench 4 (Vals AI) | 6.1% | Not measured | Only Gemini 3.5 Flash measured |
| Vibe Code Bench (Vals AI) | 48.7% | Not measured | Only Gemini 3.5 Flash measured |
| Code migration (Vals AI) | 26.8% | Not measured | Only Gemini 3.5 Flash measured |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Gemini 3.5 Flash measured |
| Terminal-Bench 2.1 (Vals AI) | 74.2% | Not measured | Only Gemini 3.5 Flash measured |
| SkillsBench (Vals AI) | 52.7% | Not measured | Only Gemini 3.5 Flash measured |
| LiveCodeBench (Artificial Analysis run) | Not measured | 23.1% | Only Phi 4 measured |
| SciCode | 53.9% | Not measured | Only Gemini 3.5 Flash measured |
| Terminal-Bench Hard (Artificial Analysis run) | 46.2% | 3.8% | Gemini 3.5 Flash |
| WeirdML | 62.6% | Not measured | Only Gemini 3.5 Flash measured |
| ALE-Bench | 911.0 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Text: coding | 1493 | 1231 | Gemini 3.5 Flash |
| LMArena Text: software and IT | 1494 | 1228 | Gemini 3.5 Flash |
| LMArena WebDev: React | 1495 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena WebDev: HTML | 1522 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena WebDev: from an image | 1504 | Not measured | Only Gemini 3.5 Flash measured |
| Neither model has a result on the other 19 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Gemini 3.5 Flash | Phi 4 | Leader |
|---|---|---|---|
| AA Intelligence Index | 33.6 | 5.9 | Gemini 3.5 Flash |
| Epoch Capabilities Index | 154.9 | 130.5 | Gemini 3.5 Flash |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Gemini 3.5 Flash | Phi 4 | Leader |
|---|---|---|---|
| Design Arena: agents building web apps | 1202 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: agents building full-stack apps | 1197 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: agents building mobile apps | 1175 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: agents building Android apps | 1190 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: agents building Godot games | 1098 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: agents building games | 1180 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: HTML slides | 1142 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: PowerPoint slides | 1179 | Not measured | Only Gemini 3.5 Flash measured |
| Finance Agent (Vals AI) | 57.9% | Not measured | Only Gemini 3.5 Flash measured |
| Tax Agent Bench (Vals AI) | 21.6% | Not measured | Only Gemini 3.5 Flash measured |
| Legal Research Bench (Vals AI) | 30.8% | Not measured | Only Gemini 3.5 Flash measured |
| Excel Modeling Benchmark (Vals AI) | 63.5% | Not measured | Only Gemini 3.5 Flash measured |
| Terminal-Bench-Science (Vals AI) | 5.7% | Not measured | Only Gemini 3.5 Flash measured |
| τ²-Bench Telecom | 95.6% | 0.0% | Gemini 3.5 Flash |
| APEX-Agents | 27.5% | Not measured | Only Gemini 3.5 Flash measured |
| BALROG | Not measured | 11.6% | Only Phi 4 measured |
| Vending-Bench 2 | $5,396 | Not measured | Only Gemini 3.5 Flash measured |
| Neither model has a result on the other 22 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index60
| Benchmark | Gemini 3.5 Flash | Phi 4 | Leader |
|---|---|---|---|
| GPQA Diamond | 92.7% | Not measured | Only Gemini 3.5 Flash measured |
| MMMU Pro (Vals AI) | 88.3% | Not measured | Only Gemini 3.5 Flash measured |
| GPQA Diamond (Artificial Analysis run) | 92.2% | 57.5% | Gemini 3.5 Flash |
| Humanity's Last Exam (Artificial Analysis run) | 42.7% | 3.8% | Gemini 3.5 Flash |
| GPQA Diamond (Epoch AI run) | 92.8% | 56.1% | Gemini 3.5 Flash |
| ARC-AGI | 92.5% | Not measured | Only Gemini 3.5 Flash measured |
| ARC-AGI-2 | 72.1% | Not measured | Only Gemini 3.5 Flash measured |
| SimpleBench | 76.7% | Not measured | Only Gemini 3.5 Flash measured |
| Chess puzzles | 45.0% | Not measured | Only Gemini 3.5 Flash measured |
| Mystery game puzzles | 32.0% | Not measured | Only Gemini 3.5 Flash measured |
| EBR-bench | 4.8% | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Text: hard prompts | 1488 | 1220 | Gemini 3.5 Flash |
| LMArena Text: expert | 1495 | 1203 | Gemini 3.5 Flash |
| LMArena Text: hard prompts in English | 1483 | 1230 | Gemini 3.5 Flash |
| LMArena Vision: diagrams | 1323 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Vision: homework | 1330 | Not measured | Only Gemini 3.5 Flash measured |
| Neither model has a result on the other 5 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Gemini 3.5 Flash | Phi 4 | Leader |
|---|---|---|---|
| AA Math Index | Not measured | 18.0 | Only Phi 4 measured |
| AIME 2025 (Artificial Analysis run) | Not measured | 18.0% | Only Phi 4 measured |
| OTIS Mock AIME | 95.6% | 13.8% | Gemini 3.5 Flash |
| FrontierMath, tiers 1 to 3 | 62.8% | Not measured | Only Gemini 3.5 Flash measured |
| FrontierMath, tier 4 | 26.8% | Not measured | Only Gemini 3.5 Flash measured |
| ProofBench | 31.0% | Not measured | Only Gemini 3.5 Flash measured |
| MATH level 5 | Not measured | 64.9% | Only Phi 4 measured |
| LMArena Text: math | 1504 | 1246 | Gemini 3.5 Flash |
| LMArena Text: mathematical work | 1495 | 1254 | Gemini 3.5 Flash |
| Neither model has a result on the other 4 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index40
| Benchmark | Gemini 3.5 Flash | Phi 4 | Leader |
|---|---|---|---|
| MMLU Pro | 89.5% | Not measured | Only Gemini 3.5 Flash measured |
| LegalBench (Vals AI) | 83.6% | Not measured | Only Gemini 3.5 Flash measured |
| CorpFin (Vals AI) | 64.7% | Not measured | Only Gemini 3.5 Flash measured |
| TaxEval (Vals AI) | 74.4% | Not measured | Only Gemini 3.5 Flash measured |
| MortgageTax (Vals AI) | 68.1% | Not measured | Only Gemini 3.5 Flash measured |
| MedCode (Vals AI) | 55.8% | Not measured | Only Gemini 3.5 Flash measured |
| MMLU Pro (Artificial Analysis run) | Not measured | 71.4% | Only Phi 4 measured |
| SimpleQA Verified | 66.2% | Not measured | Only Gemini 3.5 Flash measured |
| MMLU | Not measured | 84.8% | Only Phi 4 measured |
| LMArena Text: science | 1495 | 1206 | Gemini 3.5 Flash |
| LMArena Text: medicine and healthcare | 1483 | 1202 | Gemini 3.5 Flash |
| LMArena Text: legal and government | 1488 | 1241 | Gemini 3.5 Flash |
| LMArena Text: business and finance | 1467 | 1199 | Gemini 3.5 Flash |
| Neither model has a result on the other 4 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
| Benchmark | Gemini 3.5 Flash | Phi 4 | Leader |
|---|---|---|---|
| Design Arena: websites | 1268 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: UI components | 1273 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: game dev | 1263 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: data viz | 1236 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: 3D | 1244 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: all code categories | 1266 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: SVG | 1254 | Not measured | Only Gemini 3.5 Flash measured |
| Design Arena: ASCII art | 1254 | Not measured | Only Gemini 3.5 Flash measured |
| Neither model has a result on the other 1 board here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index180
| Benchmark | Gemini 3.5 Flash | Phi 4 | Leader |
|---|---|---|---|
| Public Benefits Bench (Vals AI) | 59.5% | Not measured | Only Gemini 3.5 Flash measured |
| SAGE (Vals AI) | 49.9% | Not measured | Only Gemini 3.5 Flash measured |
| MedScribe (Vals AI) | 76.6% | Not measured | Only Gemini 3.5 Flash measured |
| Vals Multimodal Index | 62.9% | Not measured | Only Gemini 3.5 Flash measured |
| IFBench | 76.3% | 23.5% | Gemini 3.5 Flash |
| Long context reasoning (AA-LCR) | 74.3% | 0.0% | Gemini 3.5 Flash |
| DTBench | 94.7% | Not measured | Only Gemini 3.5 Flash measured |
| LMCA | 47.1% | Not measured | Only Gemini 3.5 Flash measured |
| Creative writing (Lech Mazur) | Not measured | 62.6% | Only Phi 4 measured |
| Surface Evolver Bench | 58.1% | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Text | 1482 | 1217 | Gemini 3.5 Flash |
| LMArena Text: instruction following | 1467 | 1201 | Gemini 3.5 Flash |
| LMArena Text: creative writing | 1470 | 1181 | Gemini 3.5 Flash |
| LMArena Vision | 1308 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Text: English | 1481 | 1231 | Gemini 3.5 Flash |
| LMArena Text: non-English | 1475 | 1197 | Gemini 3.5 Flash |
| LMArena Text: Chinese | 1527 | 1212 | Gemini 3.5 Flash |
| LMArena Text: German | 1494 | 1222 | Gemini 3.5 Flash |
| LMArena Text: Japanese | 1486 | 1158 | Gemini 3.5 Flash |
| LMArena Text: Korean | 1448 | 1151 | Gemini 3.5 Flash |
| LMArena Text: French | 1490 | 1224 | Gemini 3.5 Flash |
| LMArena Text: Spanish | 1480 | 1234 | Gemini 3.5 Flash |
| LMArena Text: Russian | 1493 | 1209 | Gemini 3.5 Flash |
| LMArena Text: Polish | 1516 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Text: multi-turn | 1481 | 1206 | Gemini 3.5 Flash |
| LMArena Text: longer queries | 1483 | 1217 | Gemini 3.5 Flash |
| LMArena Text: writing and language | 1478 | 1198 | Gemini 3.5 Flash |
| LMArena Text: entertainment and media | 1459 | 1162 | Gemini 3.5 Flash |
| LMArena Vision: English | 1305 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Vision: Chinese | 1377 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Vision: reading text | 1317 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Vision: humor | 1321 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Vision: creative writing | 1301 | Not measured | Only Gemini 3.5 Flash measured |
| LMArena Document | 1463 | Not measured | Only Gemini 3.5 Flash measured |
| Neither model has a result on the other 12 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Gemini 3.5 Flash | Phi 4 | 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 3.5 Flash$0.0060Fits in one request
Phi 4$0.0001Fits in one request
Phi 4 costs less
Repository review
50K input + 3K output tokens
Gemini 3.5 Flash$0.102Fits in one request
Phi 4$0.0039Exceeds the context window
Phi 4 costs less
Agent loop
200K input + 10K output tokens
Gemini 3.5 Flash$0.390Fits in one request
Phi 4$0.015Exceeds the context window
Phi 4 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 3.5 Flash | Microsoft Phi 4 |
|---|---|---|
| Maker | Microsoft | |
| Weights | No public weights listed | Open weights |
| Released | May 19, 2026 | January 10, 2025 |
| Context window | 1.05M | 16K |
| Input $/Mtok | $1.50 | $0.07 |
| Output $/Mtok | $9.00 | $0.14 |
| Coding index | 39.8 | Not benchmarked |
| Independent measurements behind it | 4 | 0 |
| Cost per solved task | $1.21 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Gemini 3.5 Flash better than Phi 4 for coding?
Only Gemini 3.5 Flash has a coding score, 39.8 on the Vibe Coding Index; Phi 4 has none, so the two cannot be ranked on coding. Phi 4 has the edge where output tokens cost 64.3x less.
Which is cheaper, Gemini 3.5 Flash or Phi 4?
Gemini 3.5 Flash costs $1.50 per million input tokens and $9.00 per million output tokens; Phi 4 costs $0.07 and $0.14. Cost per solved task is known for Gemini 3.5 Flash only.
Which has the bigger context window, Gemini 3.5 Flash or Phi 4?
Gemini 3.5 Flash, with 1.05M tokens against 16K.
How many benchmarks is the Gemini 3.5 Flash vs Phi 4 comparison based on?
37 benchmarks have a result for both models as of October 21, 2026. Gemini 3.5 Flash's coding index rests on 4 independent benchmarks and Phi 4's on 0. Where only one model has a result, the page says so instead of naming a winner.