Grok 4.7 vs Phi 4
Updated October 21, 2026. 33 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 33 shared
Decision reading
Only Grok 4.7 has a coding score, 66.7 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 42.9x less.
Model AGrok 4.7xAI66.7/1005 independent benchmarksNo cost per solved taskModel BPhi 4MicrosoftNot benchmarkedNo coding benchmarkNo cost per solved task
- Shared results
- 33
- Grok 4.7 only
- 66
- Phi 4 only
- 12
- 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.
Coding
Vibe Coding Index, six public coding benchmarks
No call
Only Grok 4.7 has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
Grok 4.7
Grok 4.7 leads, 46.4 against 5.9.
Same board, both measured
Web design
Design Arena: websites
No call
Only Grok 4.7 has a Design Arena: websites result, 1254.
Directional only
General
Vals Index
No call
Only Grok 4.7 has a Vals Index result, 58.6%.
Directional only
Long documents
Listed context window
Grok 4.7
Grok 4.7 takes 500K tokens against 16K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Phi 4
Grok 4.7 $0.0050, Phi 4 $0.0001 per turn.
List prices
At a glance
- Coding index
- Grok 4.7 66.7 from 5 benchmarks, Phi 4 not benchmarked.
- Widest category
- Math, 47.5 points apart on average over 3 shared benchmarks.
- Widest single gap
- OTIS Mock AIME: Grok 4.7 98.1% against Phi 4 13.8%, 84.3 points apart on a 0 to 100 scale.
- Output price
- Grok 4.7 $6.00 against Phi 4 $0.14 per million output tokens, 42.9x apart.
- Context window
- Grok 4.7 takes 500K 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 | Grok 4.7 | Bars | Phi 4 | Ties |
|---|---|---|---|---|
| Coding | 2 | 0 | ||
| Intelligence | 2 | 0 | ||
| Reasoning | 5 | 0 | ||
| Math | 3 | 0 | ||
| Knowledge | 4 | 0 | ||
| General | 17 | 0 | ||
| All shared boards | 33 | 0 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index20
| Benchmark | Grok 4.7 | Phi 4 | Leader |
|---|---|---|---|
| LiveBench coding | 84.5% | Not measured | Only Grok 4.7 measured |
| LMArena WebDev | 1637 | Not measured | Only Grok 4.7 measured |
| Terminal-Bench 4 (Vals AI) | 28.8% | Not measured | Only Grok 4.7 measured |
| Vibe Code Bench (Vals AI) | 86.2% | Not measured | Only Grok 4.7 measured |
| Code migration (Vals AI) | 44.8% | Not measured | Only Grok 4.7 measured |
| ProgramBench (Vals AI) | 0.5% | Not measured | Only Grok 4.7 measured |
| IOI (Vals AI) | 57.7% | Not measured | Only Grok 4.7 measured |
| Terminal-Bench 2.1 (Vals AI) | 73.4% | Not measured | Only Grok 4.7 measured |
| SWE-bench Multimodal (Vals AI) | 42.9% | Not measured | Only Grok 4.7 measured |
| CyberBench (Vals AI) | 69.5% | Not measured | Only Grok 4.7 measured |
| LiveCodeBench (Artificial Analysis run) | Not measured | 23.1% | Only Phi 4 measured |
| SciCode | 57.8% | Not measured | Only Grok 4.7 measured |
| Terminal-Bench Hard (Artificial Analysis run) | Not measured | 3.8% | Only Phi 4 measured |
| FrontierCode | 47.6% | Not measured | Only Grok 4.7 measured |
| FrontierSWE | 29.5% | Not measured | Only Grok 4.7 measured |
| LMArena Text: coding | 1432 | 1231 | Grok 4.7 |
| LMArena Text: software and IT | 1421 | 1228 | Grok 4.7 |
| LMArena WebDev: React | 1663 | Not measured | Only Grok 4.7 measured |
| LMArena WebDev: HTML | 1627 | Not measured | Only Grok 4.7 measured |
| LMArena WebDev: from an image | 1586 | Not measured | Only Grok 4.7 measured |
| Neither model has a result on the other 18 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Grok 4.7 | Phi 4 | Leader |
|---|---|---|---|
| AA Intelligence Index | 46.4 | 5.9 | Grok 4.7 |
| Epoch Capabilities Index | 154.4 | 130.5 | Grok 4.7 |
Agentic workDirectional onlyPublic boards, shown as published, kept out of the index17 boards, none shared
| Benchmark | Grok 4.7 | Phi 4 | Leader |
|---|---|---|---|
| LiveBench research | 86.4% | Not measured | Only Grok 4.7 measured |
| LiveBench browser | 60.0% | Not measured | Only Grok 4.7 measured |
| LiveBench long agentic | 50.1% | Not measured | Only Grok 4.7 measured |
| SREBench (Vals AI) | 3.8% | Not measured | Only Grok 4.7 measured |
| Finance Agent (Vals AI) | 52.3% | Not measured | Only Grok 4.7 measured |
| Tax Agent Bench (Vals AI) | 34.2% | Not measured | Only Grok 4.7 measured |
| Legal Research Bench (Vals AI) | 47.1% | Not measured | Only Grok 4.7 measured |
| Harvey's Legal Agent Benchmark (Vals AI) | 18.3% | Not measured | Only Grok 4.7 measured |
| Excel Modeling Benchmark (Vals AI) | 67.0% | Not measured | Only Grok 4.7 measured |
| BioMysteryBench (Vals AI) | 69.3% | Not measured | Only Grok 4.7 measured |
| Terminal-Bench-Science (Vals AI) | 10.0% | Not measured | Only Grok 4.7 measured |
| Vals Science Index | 34.8% | Not measured | Only Grok 4.7 measured |
| Robo-bench (Vals AI) | 0.2% | Not measured | Only Grok 4.7 measured |
| τ²-Bench Telecom | Not measured | 0.0% | Only Phi 4 measured |
| APEX-Agents | 54.6% | Not measured | Only Grok 4.7 measured |
| BALROG | Not measured | 11.6% | Only Phi 4 measured |
| Vending-Bench 2 | $10,537 | Not measured | Only Grok 4.7 measured |
| Neither model has a result on the other 22 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index50
| Benchmark | Grok 4.7 | Phi 4 | Leader |
|---|---|---|---|
| LiveBench reasoning | 82.7% | Not measured | Only Grok 4.7 measured |
| MysteryMechanism (Vals AI) | 25.2% | Not measured | Only Grok 4.7 measured |
| GPQA Diamond (Artificial Analysis run) | Not measured | 57.5% | Only Phi 4 measured |
| Humanity's Last Exam (Artificial Analysis run) | 43.1% | 3.8% | Grok 4.7 |
| GPQA Diamond (Epoch AI run) | 92.7% | 56.1% | Grok 4.7 |
| ARC-AGI | 90.2% | Not measured | Only Grok 4.7 measured |
| ARC-AGI-2 | 61.4% | Not measured | Only Grok 4.7 measured |
| Chess puzzles | 41.0% | Not measured | Only Grok 4.7 measured |
| Mystery game puzzles | 29.0% | Not measured | Only Grok 4.7 measured |
| LMArena Text: hard prompts | 1415 | 1220 | Grok 4.7 |
| LMArena Text: expert | 1439 | 1203 | Grok 4.7 |
| LMArena Text: hard prompts in English | 1427 | 1230 | Grok 4.7 |
| LMArena Vision: diagrams | 1254 | Not measured | Only Grok 4.7 measured |
| LMArena Vision: homework | 1309 | Not measured | Only Grok 4.7 measured |
| Neither model has a result on the other 7 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Grok 4.7 | 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 | 98.1% | 13.8% | Grok 4.7 |
| FrontierMath, tiers 1 to 3 | 53.0% | Not measured | Only Grok 4.7 measured |
| FrontierMath, tier 4 | 17.1% | Not measured | Only Grok 4.7 measured |
| ProofBench | 26.0% | Not measured | Only Grok 4.7 measured |
| MATH level 5 | Not measured | 64.9% | Only Phi 4 measured |
| LMArena Text: math | 1414 | 1246 | Grok 4.7 |
| LMArena Text: mathematical work | 1435 | 1254 | Grok 4.7 |
| Neither model has a result on the other 4 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index40
| Benchmark | Grok 4.7 | Phi 4 | Leader |
|---|---|---|---|
| LegalBench (Vals AI) | 84.4% | Not measured | Only Grok 4.7 measured |
| MedCode (Vals AI) | 49.5% | Not measured | Only Grok 4.7 measured |
| MMLU Pro (Artificial Analysis run) | Not measured | 71.4% | Only Phi 4 measured |
| SimpleQA Verified | 56.0% | Not measured | Only Grok 4.7 measured |
| MMLU | Not measured | 84.8% | Only Phi 4 measured |
| LMArena Text: science | 1419 | 1206 | Grok 4.7 |
| LMArena Text: medicine and healthcare | 1395 | 1202 | Grok 4.7 |
| LMArena Text: legal and government | 1423 | 1241 | Grok 4.7 |
| LMArena Text: business and finance | 1389 | 1199 | Grok 4.7 |
| Neither model has a result on the other 8 boards here | |||
Web designDirectional onlyPublic boards, shown as published, kept out of the index7 boards, none shared
| Benchmark | Grok 4.7 | Phi 4 | Leader |
|---|---|---|---|
| Design Arena: websites | 1254 | Not measured | Only Grok 4.7 measured |
| Design Arena: UI components | 1273 | Not measured | Only Grok 4.7 measured |
| Design Arena: game dev | 1312 | Not measured | Only Grok 4.7 measured |
| Design Arena: data viz | 1235 | Not measured | Only Grok 4.7 measured |
| Design Arena: 3D | 1270 | Not measured | Only Grok 4.7 measured |
| Design Arena: all code categories | 1271 | Not measured | Only Grok 4.7 measured |
| LiveBench design | 78.0% | Not measured | Only Grok 4.7 measured |
| Neither model has a result on the other 2 boards here | |||
GeneralLike-for-likePublic boards, shown as published, kept out of the index170
| Benchmark | Grok 4.7 | Phi 4 | Leader |
|---|---|---|---|
| Vals Index | 58.6% | Not measured | Only Grok 4.7 measured |
| LiveBench data analysis | 76.9% | Not measured | Only Grok 4.7 measured |
| Public Benefits Bench (Vals AI) | 65.6% | Not measured | Only Grok 4.7 measured |
| SAGE (Vals AI) | 40.8% | Not measured | Only Grok 4.7 measured |
| MedScribe (Vals AI) | 89.4% | Not measured | Only Grok 4.7 measured |
| IFBench | Not measured | 23.5% | Only Phi 4 measured |
| Long context reasoning (AA-LCR) | 80.0% | 0.0% | Grok 4.7 |
| DTBench | 96.0% | Not measured | Only Grok 4.7 measured |
| LMCA | 49.4% | Not measured | Only Grok 4.7 measured |
| Creative writing (Lech Mazur) | Not measured | 62.6% | Only Phi 4 measured |
| Furniture assembly | 20.8% | Not measured | Only Grok 4.7 measured |
| LMArena Text | 1399 | 1217 | Grok 4.7 |
| LMArena Text: instruction following | 1406 | 1201 | Grok 4.7 |
| LMArena Text: creative writing | 1381 | 1181 | Grok 4.7 |
| LMArena Vision | 1231 | Not measured | Only Grok 4.7 measured |
| LMArena Text: English | 1411 | 1231 | Grok 4.7 |
| LMArena Text: non-English | 1386 | 1197 | Grok 4.7 |
| LMArena Text: Chinese | 1449 | 1212 | Grok 4.7 |
| LMArena Text: German | 1366 | 1222 | Grok 4.7 |
| LMArena Text: Japanese | 1359 | 1158 | Grok 4.7 |
| LMArena Text: Korean | 1383 | 1151 | Grok 4.7 |
| LMArena Text: French | 1431 | 1224 | Grok 4.7 |
| LMArena Text: Spanish | 1406 | 1234 | Grok 4.7 |
| LMArena Text: Russian | 1399 | 1209 | Grok 4.7 |
| LMArena Text: Polish | 1420 | Not measured | Only Grok 4.7 measured |
| LMArena Text: multi-turn | 1394 | 1206 | Grok 4.7 |
| LMArena Text: longer queries | 1416 | 1217 | Grok 4.7 |
| LMArena Text: writing and language | 1394 | 1198 | Grok 4.7 |
| LMArena Text: entertainment and media | 1376 | 1162 | Grok 4.7 |
| LMArena Vision: English | 1229 | Not measured | Only Grok 4.7 measured |
| LMArena Vision: Chinese | 1250 | Not measured | Only Grok 4.7 measured |
| LMArena Vision: reading text | 1241 | Not measured | Only Grok 4.7 measured |
| LMArena Vision: creative writing | 1217 | Not measured | Only Grok 4.7 measured |
| 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
| Benchmark | Grok 4.7 | 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
Grok 4.7$0.0050Fits in one request
Phi 4$0.0001Fits in one request
Phi 4 costs less
Repository review
50K input + 3K output tokens
Grok 4.7$0.118Fits in one request
Phi 4$0.0039Exceeds the context window
Phi 4 costs less
Agent loop
200K input + 10K output tokens
Grok 4.7$0.460Fits 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 | SpaceXAI Grok 4.7 | Microsoft Phi 4 |
|---|---|---|
| Maker | xAI | Microsoft |
| Weights | No public weights listed | Open weights |
| Released | September 21, 2026 | January 10, 2025 |
| Context window | 500K | 16K |
| Input $/Mtok | $2.00 | $0.07 |
| Output $/Mtok | $6.00 | $0.14 |
| Coding index | 66.7 | Not benchmarked |
| Independent measurements behind it | 5 | 0 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Grok 4.7 better than Phi 4 for coding?
Only Grok 4.7 has a coding score, 66.7 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 42.9x less.
Which is cheaper, Grok 4.7 or Phi 4?
Grok 4.7 costs $2.00 per million input tokens and $6.00 per million output tokens; Phi 4 costs $0.07 and $0.14. Cost per solved task is known for neither of them.
Which has the bigger context window, Grok 4.7 or Phi 4?
Grok 4.7, with 500K tokens against 16K.
How many benchmarks is the Grok 4.7 vs Phi 4 comparison based on?
33 benchmarks have a result for both models as of October 21, 2026. Grok 4.7's coding index rests on 5 independent benchmarks and Phi 4's on 0. Where only one model has a result, the page says so instead of naming a winner.