Grok 4.5 vs Mistral Large 3 2512
Grok 4.5 leads Mistral Large 3 2512 on the Artificial Analysis Intelligence Index, 38.8 to 9.3, and Mistral Large 3 2512 costs 4.0x less per output token ($1.50 vs $6.00 per million). Updated October 22, 2026; 24 benchmarks have a result for both.
Grok 4.5 or Mistral Large 3 2512: which to pick
Only Grok 4.5 has a coding score, 58.6 on the Vibe Coding Index; Mistral Large 3 2512 has none, so the two cannot be ranked on coding.
Pick Mistral Large 3 2512 for output tokens 4.0x cheaper ($1.50 vs $6.00 per million) and work measured by MMMU Pro (Vals AI) or TaxEval (Vals AI), where it leads.
- Shared results
- 24
- Grok 4.5 only
- 100
- Mistral Large 3 2512 only
- 4
- 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
No call
Only Grok 4.5 has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
Grok 4.5
Grok 4.5 leads, 38.8 against 9.3.
Same board, both measured
Agentic work
AA Agentic Index
Grok 4.5
Grok 4.5 leads, 41.2 against 1.0.
Same board, both measured
Reasoning
GPQA Diamond
Grok 4.5
Grok 4.5 leads, 92.9% against 68.4%.
Same board, both measured
Math
AIME
No call
Only Mistral Large 3 2512 has a AIME result, 42.9%.
Directional only
Knowledge
MMLU Pro
Grok 4.5
Grok 4.5 leads, 89.2% against 79.8%.
Same board, both measured
Web design
Design Arena: websites
Grok 4.5
Grok 4.5 leads, 1281 against 1162.
Same board, both measured
Long documents
Listed context window
Grok 4.5
Grok 4.5 takes 500K tokens against 262K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Mistral Large 3 2512
Grok 4.5 $0.0050, Mistral Large 3 2512 $0.0013 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
Grok 4.5
Grok 4.5 $0.62, Mistral Large 3 2512 $0.77.
Published attempt cost and pass rate
At a glance
- Coding index
- Grok 4.5 58.6 from 6 benchmarks, Mistral Large 3 2512 not benchmarked.
- Widest category
- Coding, 41.1 points apart on average over 4 shared benchmarks.
- Widest single gap
- AA Coding Index: Grok 4.5 72.4 against Mistral Large 3 2512 20.1, 52.3 points apart on a 0 to 100 scale.
- Output price
- Grok 4.5 $6.00 against Mistral Large 3 2512 $1.50 per million output tokens, 4.0x apart.
- Context window
- Grok 4.5 takes 500K tokens, Mistral Large 3 2512 262K.
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 4.5 | Bars | Mistral Large 3 2512 | Ties |
|---|---|---|---|---|
| Coding | 4 | 0 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 1 | 0 | ||
| Reasoning | 1 | 1 | ||
| Knowledge | 4 | 1 | ||
| Web design | 8 | 0 | ||
| General | 3 | 0 | ||
| All shared boards | 22 | 2 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index40
| Benchmark | Grok 4.5 | Mistral Large 3 2512 | Leader |
|---|---|---|---|
| LiveBench coding | 94.3% | Not measured | Only Grok 4.5 measured |
| LMArena WebDev | 1553 | Not measured | Only Grok 4.5 measured |
| AA Coding Index | 72.4 | 20.1 | Grok 4.5 |
| LiveCodeBench (Vals AI) | 87.3% | 55.3% | Grok 4.5 |
| SWE-bench Verified (Vals AI) | 86.6% | 41.4% | Grok 4.5 |
| Terminal-Bench 4 (Vals AI) | 8.6% | Not measured | Only Grok 4.5 measured |
| Vibe Code Bench (Vals AI) | 69.0% | Not measured | Only Grok 4.5 measured |
| Code migration (Vals AI) | 36.6% | Not measured | Only Grok 4.5 measured |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Grok 4.5 measured |
| IOI (Vals AI) | 40.6% | Not measured | Only Grok 4.5 measured |
| Terminal-Bench 2.1 (Vals AI) | 67.8% | Not measured | Only Grok 4.5 measured |
| SkillsBench (Vals AI) | 66.0% | Not measured | Only Grok 4.5 measured |
| CyberBench (Vals AI) | 71.0% | Not measured | Only Grok 4.5 measured |
| SciCode | 55.0% | Not measured | Only Grok 4.5 measured |
| FrontierCode | 42.4% | Not measured | Only Grok 4.5 measured |
| WeirdML | 46.4% | Not measured | Only Grok 4.5 measured |
| ALE-Bench | 1309.3 | 264.7 | Grok 4.5 |
| LMArena Text: coding | 1474 | Not measured | Only Grok 4.5 measured |
| LMArena Text: software and IT | 1471 | Not measured | Only Grok 4.5 measured |
| LMArena WebDev: React | 1556 | Not measured | Only Grok 4.5 measured |
| LMArena WebDev: HTML | 1567 | Not measured | Only Grok 4.5 measured |
| LMArena WebDev: from an image | 1562 | Not measured | Only Grok 4.5 measured |
| SciCode | 6.5% | Not measured | Only Grok 4.5 measured |
| Neither model has a result on the other 20 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Grok 4.5 | Mistral Large 3 2512 | Leader |
|---|---|---|---|
| AA Intelligence Index | 38.8 | 9.3 | Grok 4.5 |
| Epoch Capabilities Index | 154.1 | Not measured | Only Grok 4.5 measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Grok 4.5 | Mistral Large 3 2512 | Leader |
|---|---|---|---|
| AA Agentic Index | 41.2 | 1.0 | Grok 4.5 |
| Design Arena: agents building web apps | 1218 | Not measured | Only Grok 4.5 measured |
| Design Arena: agents building full-stack apps | 1249 | Not measured | Only Grok 4.5 measured |
| Design Arena: agents building mobile apps | 1220 | Not measured | Only Grok 4.5 measured |
| Design Arena: agents building Android apps | 1236 | Not measured | Only Grok 4.5 measured |
| Design Arena: agents building Godot games | 1229 | Not measured | Only Grok 4.5 measured |
| Design Arena: agents building games | 1236 | Not measured | Only Grok 4.5 measured |
| Design Arena: HTML slides | 1155 | Not measured | Only Grok 4.5 measured |
| Design Arena: PowerPoint slides | 1209 | Not measured | Only Grok 4.5 measured |
| LiveBench research | 85.9% | Not measured | Only Grok 4.5 measured |
| LiveBench browser | 44.0% | Not measured | Only Grok 4.5 measured |
| LiveBench long agentic | 27.9% | Not measured | Only Grok 4.5 measured |
| SREBench (Vals AI) | 0.8% | Not measured | Only Grok 4.5 measured |
| Finance Agent (Vals AI) | 48.4% | Not measured | Only Grok 4.5 measured |
| Tax Agent Bench (Vals AI) | 24.9% | Not measured | Only Grok 4.5 measured |
| Legal Research Bench (Vals AI) | 38.0% | Not measured | Only Grok 4.5 measured |
| Excel Modeling Benchmark (Vals AI) | 52.9% | Not measured | Only Grok 4.5 measured |
| Time Horizon Index (Vals AI) | 6.3% | Not measured | Only Grok 4.5 measured |
| APEX-Agents | 56.2% | Not measured | Only Grok 4.5 measured |
| PostTrainBench | 23.4% | Not measured | Only Grok 4.5 measured |
| Vending-Bench 2 | $3,887 | Not measured | Only Grok 4.5 measured |
| LMArena Search | 1216 | Not measured | Only Grok 4.5 measured |
| Neither model has a result on the other 20 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | Grok 4.5 | Mistral Large 3 2512 | Leader |
|---|---|---|---|
| GPQA Diamond | 92.9% | 68.4% | Grok 4.5 |
| LiveBench reasoning | 87.2% | Not measured | Only Grok 4.5 measured |
| MMMU Pro (Vals AI) | 61.8% | 66.2% | Mistral Large 3 2512 |
| GPQA Diamond (Artificial Analysis run) | 93.1% | Not measured | Only Grok 4.5 measured |
| Humanity's Last Exam (Artificial Analysis run) | 42.7% | Not measured | Only Grok 4.5 measured |
| GPQA Diamond (Epoch AI run) | 93.4% | Not measured | Only Grok 4.5 measured |
| ARC-AGI | 87.2% | Not measured | Only Grok 4.5 measured |
| ARC-AGI-2 | 52.6% | Not measured | Only Grok 4.5 measured |
| SimpleBench | 70.0% | Not measured | Only Grok 4.5 measured |
| Chess puzzles | 35.0% | Not measured | Only Grok 4.5 measured |
| LMArena Text: hard prompts | 1462 | Not measured | Only Grok 4.5 measured |
| LMArena Text: expert | 1465 | Not measured | Only Grok 4.5 measured |
| LMArena Text: hard prompts in English | 1464 | Not measured | Only Grok 4.5 measured |
| LMArena Vision: diagrams | 1304 | Not measured | Only Grok 4.5 measured |
| LMArena Vision: homework | 1295 | Not measured | Only Grok 4.5 measured |
| Kagi LLM Benchmark | 83.5% | Not measured | Only Grok 4.5 measured |
| Neither model has a result on the other 7 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index9 boards, none shared
| Benchmark | Grok 4.5 | Mistral Large 3 2512 | Leader |
|---|---|---|---|
| AIME | Not measured | 42.9% | Only Mistral Large 3 2512 measured |
| MGSM (Vals AI) | Not measured | 85.4% | Only Mistral Large 3 2512 measured |
| OTIS Mock AIME | 97.8% | Not measured | Only Grok 4.5 measured |
| FrontierMath, tiers 1 to 3 | 57.2% | Not measured | Only Grok 4.5 measured |
| FrontierMath, tier 4 | 24.4% | Not measured | Only Grok 4.5 measured |
| ProofBench | 31.0% | Not measured | Only Grok 4.5 measured |
| LMArena Text: math | 1460 | Not measured | Only Grok 4.5 measured |
| LMArena Text: mathematical work | 1482 | Not measured | Only Grok 4.5 measured |
| MathArena | 34.1% | Not measured | Only Grok 4.5 measured |
| Neither model has a result on the other 5 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index41
| Benchmark | Grok 4.5 | Mistral Large 3 2512 | Leader |
|---|---|---|---|
| MMLU Pro | 89.2% | 79.8% | Grok 4.5 |
| LegalBench (Vals AI) | 86.0% | 79.1% | Grok 4.5 |
| CaseLaw (Vals AI) | Not measured | 61.4% | Only Mistral Large 3 2512 measured |
| CorpFin (Vals AI) | 67.4% | 61.0% | Grok 4.5 |
| TaxEval (Vals AI) | 71.7% | 73.1% | Mistral Large 3 2512 |
| MortgageTax (Vals AI) | 61.8% | 52.1% | Grok 4.5 |
| MedQA (Vals AI) | Not measured | 82.2% | Only Mistral Large 3 2512 measured |
| MedCode (Vals AI) | 43.3% | Not measured | Only Grok 4.5 measured |
| SimpleQA Verified | 48.3% | Not measured | Only Grok 4.5 measured |
| LMArena Text: science | 1464 | Not measured | Only Grok 4.5 measured |
| LMArena Text: medicine and healthcare | 1440 | Not measured | Only Grok 4.5 measured |
| LMArena Text: legal and government | 1463 | Not measured | Only Grok 4.5 measured |
| LMArena Text: business and finance | 1441 | Not measured | Only Grok 4.5 measured |
| FACTS Benchmark Suite | 50.2% | Not measured | Only Grok 4.5 measured |
| Neither model has a result on the other 6 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index80
| Benchmark | Grok 4.5 | Mistral Large 3 2512 | Leader |
|---|---|---|---|
| Design Arena: websites | 1281 | 1162 | Grok 4.5 |
| Design Arena: UI components | 1277 | 1099 | Grok 4.5 |
| Design Arena: game dev | 1263 | 1079 | Grok 4.5 |
| Design Arena: data viz | 1266 | 1137 | Grok 4.5 |
| Design Arena: 3D | 1257 | 1099 | Grok 4.5 |
| Design Arena: all code categories | 1275 | 1141 | Grok 4.5 |
| Design Arena: SVG | 1214 | 992 | Grok 4.5 |
| Design Arena: ASCII art | 1257 | 1069 | Grok 4.5 |
| LiveBench design | 78.1% | Not measured | Only Grok 4.5 measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Grok 4.5 | Mistral Large 3 2512 | Leader |
|---|---|---|---|
| LiveBench data analysis | 73.0% | Not measured | Only Grok 4.5 measured |
| SAGE (Vals AI) | 35.0% | 24.6% | Grok 4.5 |
| MedScribe (Vals AI) | 86.9% | Not measured | Only Grok 4.5 measured |
| Vals Multimodal Index | 63.4% | Not measured | Only Grok 4.5 measured |
| Long context reasoning (AA-LCR) | 79.3% | Not measured | Only Grok 4.5 measured |
| DTBench | 96.5% | 65.1% | Grok 4.5 |
| LMCA | 45.2% | 16.7% | Grok 4.5 |
| Surface Evolver Bench | 74.4% | Not measured | Only Grok 4.5 measured |
| Furniture assembly | 22.5% | Not measured | Only Grok 4.5 measured |
| LMArena Text | 1448 | Not measured | Only Grok 4.5 measured |
| LMArena Text: instruction following | 1445 | Not measured | Only Grok 4.5 measured |
| LMArena Text: creative writing | 1442 | Not measured | Only Grok 4.5 measured |
| LMArena Vision | 1287 | Not measured | Only Grok 4.5 measured |
| LMArena Text: English | 1453 | Not measured | Only Grok 4.5 measured |
| LMArena Text: non-English | 1440 | Not measured | Only Grok 4.5 measured |
| LMArena Text: Chinese | 1495 | Not measured | Only Grok 4.5 measured |
| LMArena Text: German | 1448 | Not measured | Only Grok 4.5 measured |
| LMArena Text: Japanese | 1430 | Not measured | Only Grok 4.5 measured |
| LMArena Text: Korean | 1406 | Not measured | Only Grok 4.5 measured |
| LMArena Text: French | 1457 | Not measured | Only Grok 4.5 measured |
| LMArena Text: Spanish | 1451 | Not measured | Only Grok 4.5 measured |
| LMArena Text: Russian | 1449 | Not measured | Only Grok 4.5 measured |
| LMArena Text: Polish | 1432 | Not measured | Only Grok 4.5 measured |
| LMArena Text: multi-turn | 1456 | Not measured | Only Grok 4.5 measured |
| LMArena Text: longer queries | 1462 | Not measured | Only Grok 4.5 measured |
| LMArena Text: writing and language | 1445 | Not measured | Only Grok 4.5 measured |
| LMArena Text: entertainment and media | 1434 | Not measured | Only Grok 4.5 measured |
| LMArena Vision: English | 1285 | Not measured | Only Grok 4.5 measured |
| LMArena Vision: Chinese | 1345 | Not measured | Only Grok 4.5 measured |
| LMArena Vision: reading text | 1293 | Not measured | Only Grok 4.5 measured |
| LMArena Vision: humor | 1276 | Not measured | Only Grok 4.5 measured |
| LMArena Vision: creative writing | 1293 | Not measured | Only Grok 4.5 measured |
| LMArena Document | 1452 | Not measured | Only Grok 4.5 measured |
| Neither model has a result on the other 15 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Grok 4.5 | Mistral Large 3 2512 | 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
Mistral Large 3 2512 costs less
Repository review
50K input + 3K output tokens
Mistral Large 3 2512 costs less
Agent loop
200K input + 10K output tokens
Mistral Large 3 2512 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 4.5 | Mistral Large 3 2512 |
|---|---|---|
| Maker | xAI | Mistral AI |
| Weights | No public weights listed | No public weights listed |
| Released | July 8, 2026 | December 1, 2025 |
| Context window | 500K | 262K |
| Input $/Mtok | $2.00 | $0.50 |
| Output $/Mtok | $6.00 | $1.50 |
| Coding index | 58.6 | Not benchmarked |
| Independent measurements behind it | 6 | 1 |
| Cost per solved task | $0.62 | $0.77 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Grok 4.5 better than Mistral Large 3 2512 for coding?
Only Grok 4.5 has a coding score, 58.6 on the Vibe Coding Index; Mistral Large 3 2512 has none, so the two cannot be ranked on coding. Pick Mistral Large 3 2512 for output tokens 4.0x cheaper ($1.50 vs $6.00 per million) and work measured by MMMU Pro (Vals AI) or TaxEval (Vals AI), where it leads.
Which is cheaper, Grok 4.5 or Mistral Large 3 2512?
Grok 4.5 costs $2.00 per million input tokens and $6.00 per million output tokens; Mistral Large 3 2512 costs $0.50 and $1.50. Per solved SWE-bench Verified task, Grok 4.5 costs $0.62 and Mistral Large 3 2512 $0.77.
Which has the bigger context window, Grok 4.5 or Mistral Large 3 2512?
Grok 4.5, with 500K tokens against 262K.
How many benchmarks is the Grok 4.5 vs Mistral Large 3 2512 comparison based on?
24 benchmarks have a result for both models as of October 22, 2026. Grok 4.5's coding index rests on 6 independent benchmarks and Mistral Large 3 2512's on 1. Where only one model has a result, the page says so instead of naming a winner.