GPT-6 Astra vs Mistral Large 4
GPT-6 Astra leads Mistral Large 4 on coding, 85.1 to 56.4 on the Vibe Coding Index, and Mistral Large 4 costs 12.0x less per output token ($4.18 vs $50.00 per million). Updated October 22, 2026; 80 benchmarks have a result for both.
GPT-6 Astra or Mistral Large 4: which to pick
Pick GPT-6 Astra for coding: it scores 85.1 on the Vibe Coding Index against 56.4 for Mistral Large 4.
Pick Mistral Large 4 for output tokens 12.0x cheaper ($4.18 vs $50.00 per million) and work measured by Finance Agent (Vals AI) or Tax Agent Bench (Vals AI), where it leads.
- Shared results
- 80
- GPT-6 Astra only
- 41
- Mistral Large 4 only
- 1
- Like-for-like categories
- 8 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
GPT-6 Astra
GPT-6 Astra scores 85.1 against 56.4.
5 benchmarks behind each score
Intelligence
AA Intelligence Index
GPT-6 Astra
GPT-6 Astra leads, 52.7 against 38.4.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only GPT-6 Astra has a AA Agentic Index result, 51.0.
Directional only
Web design
Design Arena: websites
GPT-6 Astra
GPT-6 Astra leads, 1336 against 1257.
Same board, both measured
General
Vals Index
No call
Only GPT-6 Astra has a Vals Index result, 65.0%.
Directional only
Long documents
Listed context window
GPT-6 Astra
GPT-6 Astra takes 1.05M tokens against 1.05M.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Mistral Large 4
GPT-6 Astra $0.035, Mistral Large 4 $0.0034 per turn.
List prices
At a glance
- Coding index
- GPT-6 Astra 85.1 from 5 benchmarks, Mistral Large 4 56.4 from 5 benchmarks.
- Widest category
- Math, 33.5 points apart on average over 6 shared benchmarks.
- Widest single gap
- ProofBench: GPT-6 Astra 99.0% against Mistral Large 4 10.0%, 89.0 points apart on a 0 to 100 scale.
- Output price
- GPT-6 Astra $50.00 against Mistral Large 4 $4.18 per million output tokens, 12.0x apart.
- Context window
- GPT-6 Astra takes 1.05M tokens, Mistral Large 4 1.05M.
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 | GPT-6 Astra | Bars | Mistral Large 4 | Ties |
|---|---|---|---|---|
| Coding | 15 | 0 | ||
| Intelligence | 2 | 0 | ||
| Agentic work | 6 | 3 | ||
| Reasoning | 10 | 0 | ||
| Math | 5 | 1 | ||
| Knowledge | 5 | 1 | ||
| Web design | 9 | 0 | ||
| General | 17 | 6 | ||
| All shared boards | 69 | 11 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index150
| Benchmark | GPT-6 Astra | Mistral Large 4 | Leader |
|---|---|---|---|
| LiveBench coding | 93.7% | 60.0% | GPT-6 Astra |
| LMArena WebDev | 1774 | 1543 | GPT-6 Astra |
| AA Coding Index | 76.9 | Not measured | Only GPT-6 Astra measured |
| Terminal-Bench 4 (Vals AI) | 59.6% | 27.8% | GPT-6 Astra |
| Vibe Code Bench (Vals AI) | 89.6% | 78.4% | GPT-6 Astra |
| Code migration (Vals AI) | 67.7% | 30.6% | GPT-6 Astra |
| ProgramBench (Vals AI) | 5.5% | Not measured | Only GPT-6 Astra measured |
| IOI (Vals AI) | 100.0% | 45.3% | GPT-6 Astra |
| Terminal-Bench 2.1 (Vals AI) | 87.3% | Not measured | Only GPT-6 Astra measured |
| SWE-bench Multimodal (Vals AI) | 43.3% | 33.1% | GPT-6 Astra |
| Vibe Code Bench 1-100 (Vals AI) | 27.6% | 14.3% | GPT-6 Astra |
| CyberBench (Vals AI) | 41.1% | Not measured | Only GPT-6 Astra measured |
| SciCode | 56.5% | 54.2% | GPT-6 Astra |
| FrontierCode | 53.3% | 39.4% | GPT-6 Astra |
| FrontierSWE | 65.5% | Not measured | Only GPT-6 Astra measured |
| GSO-Bench | 79.4% | Not measured | Only GPT-6 Astra measured |
| WeirdML | 93.3% | Not measured | Only GPT-6 Astra measured |
| MirrorCode | 46.7% | Not measured | Only GPT-6 Astra measured |
| ALE-Bench | 2951.3 | 1213.2 | GPT-6 Astra |
| LMArena Text: coding | 1489 | 1478 | GPT-6 Astra |
| LMArena Text: software and IT | 1474 | 1467 | GPT-6 Astra |
| LMArena WebDev: React | 1798 | 1559 | GPT-6 Astra |
| LMArena WebDev: HTML | 1775 | 1487 | GPT-6 Astra |
| LMArena WebDev: from an image | 1726 | Not measured | Only GPT-6 Astra measured |
| SciCode | 12.3% | Not measured | Only GPT-6 Astra measured |
| Neither model has a result on the other 14 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | GPT-6 Astra | Mistral Large 4 | Leader |
|---|---|---|---|
| AA Intelligence Index | 52.7 | 38.4 | GPT-6 Astra |
| Epoch Capabilities Index | 168.8 | 150.0 | GPT-6 Astra |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index63
| Benchmark | GPT-6 Astra | Mistral Large 4 | Leader |
|---|---|---|---|
| AA Agentic Index | 51.0 | Not measured | Only GPT-6 Astra measured |
| Design Arena: agents building web apps | 1373 | Not measured | Only GPT-6 Astra measured |
| Design Arena: agents building full-stack apps | 1357 | Not measured | Only GPT-6 Astra measured |
| Design Arena: agents building mobile apps | 1312 | Not measured | Only GPT-6 Astra measured |
| Design Arena: agents building Godot games | 1348 | Not measured | Only GPT-6 Astra measured |
| Design Arena: PowerPoint slides | 1219 | Not measured | Only GPT-6 Astra measured |
| Design Arena: spreadsheets | 1365 | Not measured | Only GPT-6 Astra measured |
| LiveBench research | 85.2% | 28.1% | GPT-6 Astra |
| LiveBench browser | 84.0% | 8.0% | GPT-6 Astra |
| LiveBench long agentic | 82.4% | 11.8% | GPT-6 Astra |
| SREBench (Vals AI) | 56.9% | Not measured | Only GPT-6 Astra measured |
| Finance Agent (Vals AI) | 53.5% | 54.7% | Mistral Large 4 |
| Tax Agent Bench (Vals AI) | 20.7% | 30.4% | Mistral Large 4 |
| Legal Research Bench (Vals AI) | 39.4% | 31.7% | GPT-6 Astra |
| Harvey's Legal Agent Benchmark (Vals AI) | 7.1% | 17.5% | Mistral Large 4 |
| Excel Modeling Benchmark (Vals AI) | 71.7% | 56.2% | GPT-6 Astra |
| BioMysteryBench (Vals AI) | 79.3% | 67.0% | GPT-6 Astra |
| Terminal-Bench-Science (Vals AI) | 62.9% | Not measured | Only GPT-6 Astra measured |
| Vals Science Index | 65.1% | Not measured | Only GPT-6 Astra measured |
| Time Horizon Index (Vals AI) | 90.5% | Not measured | Only GPT-6 Astra measured |
| CUA-bench (Vals AI) | 19.2% | Not measured | Only GPT-6 Astra measured |
| Robo-bench (Vals AI) | 23.3% | Not measured | Only GPT-6 Astra measured |
| APEX-Agents | 64.7% | Not measured | Only GPT-6 Astra measured |
| Remote Labor Index | 20.8% | Not measured | Only GPT-6 Astra measured |
| BALROG | 68.3% | Not measured | Only GPT-6 Astra measured |
| SpireClimb | 25.7% | Not measured | Only GPT-6 Astra measured |
| Vending-Bench 2 | $15,515 | Not measured | Only GPT-6 Astra measured |
| Neither model has a result on the other 14 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index100
| Benchmark | GPT-6 Astra | Mistral Large 4 | Leader |
|---|---|---|---|
| LiveBench reasoning | 92.7% | 83.9% | GPT-6 Astra |
| MysteryMechanism (Vals AI) | 53.1% | 12.2% | GPT-6 Astra |
| GPQA Diamond (Artificial Analysis run) | 96.3% | Not measured | Only GPT-6 Astra measured |
| Humanity's Last Exam (Artificial Analysis run) | 54.7% | 35.0% | GPT-6 Astra |
| GPQA Diamond (Epoch AI run) | 95.8% | 88.1% | GPT-6 Astra |
| ARC-AGI | 98.5% | Not measured | Only GPT-6 Astra measured |
| ARC-AGI-2 | 95.0% | Not measured | Only GPT-6 Astra measured |
| Chess puzzles | 74.0% | 33.0% | GPT-6 Astra |
| Mystery game puzzles | 84.0% | 17.0% | GPT-6 Astra |
| EBR-bench | 76.2% | Not measured | Only GPT-6 Astra measured |
| LMArena Text: hard prompts | 1463 | 1444 | GPT-6 Astra |
| LMArena Text: expert | 1489 | 1444 | GPT-6 Astra |
| LMArena Text: hard prompts in English | 1464 | 1442 | GPT-6 Astra |
| LMArena Vision: diagrams | 1297 | 1288 | GPT-6 Astra |
| LMArena Vision: homework | 1337 | Not measured | Only GPT-6 Astra measured |
| Neither model has a result on the other 7 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index51
| Benchmark | GPT-6 Astra | Mistral Large 4 | Leader |
|---|---|---|---|
| OTIS Mock AIME | 100.0% | 98.3% | GPT-6 Astra |
| FrontierMath, tiers 1 to 3 | 93.7% | 58.6% | GPT-6 Astra |
| FrontierMath, tier 4 | 97.6% | 26.8% | GPT-6 Astra |
| ProofBench | 99.0% | 10.0% | GPT-6 Astra |
| LMArena Text: math | 1465 | 1469 | Mistral Large 4 |
| LMArena Text: mathematical work | 1484 | 1462 | GPT-6 Astra |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index51
| Benchmark | GPT-6 Astra | Mistral Large 4 | Leader |
|---|---|---|---|
| MedCode (Vals AI) | 48.5% | 40.7% | GPT-6 Astra |
| SimpleQA Verified | 75.6% | 20.0% | GPT-6 Astra |
| LMArena Text: science | 1451 | 1448 | GPT-6 Astra |
| LMArena Text: medicine and healthcare | 1425 | 1433 | Mistral Large 4 |
| LMArena Text: legal and government | 1462 | 1425 | GPT-6 Astra |
| LMArena Text: business and finance | 1439 | 1439 | GPT-6 Astra |
| FACTS Benchmark Suite | 75.7% | Not measured | Only GPT-6 Astra measured |
| Neither model has a result on the other 11 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index90
| Benchmark | GPT-6 Astra | Mistral Large 4 | Leader |
|---|---|---|---|
| Design Arena: websites | 1336 | 1257 | GPT-6 Astra |
| Design Arena: UI components | 1393 | 1272 | GPT-6 Astra |
| Design Arena: game dev | 1478 | 1225 | GPT-6 Astra |
| Design Arena: data viz | 1335 | 1275 | GPT-6 Astra |
| Design Arena: 3D | 1500 | 1252 | GPT-6 Astra |
| Design Arena: all code categories | 1379 | 1259 | GPT-6 Astra |
| Design Arena: SVG | 1431 | 1232 | GPT-6 Astra |
| Design Arena: ASCII art | 1353 | 1202 | GPT-6 Astra |
| LiveBench design | 82.0% | 65.6% | GPT-6 Astra |
GeneralLike-for-likePublic boards, shown as published, kept out of the index176
| Benchmark | GPT-6 Astra | Mistral Large 4 | Leader |
|---|---|---|---|
| Vals Index | 65.0% | Not measured | Only GPT-6 Astra measured |
| LiveBench data analysis | 83.0% | 76.5% | GPT-6 Astra |
| SAGE (Vals AI) | 46.4% | Not measured | Only GPT-6 Astra measured |
| MedScribe (Vals AI) | 87.9% | 80.4% | GPT-6 Astra |
| Long context reasoning (AA-LCR) | 80.7% | 81.3% | Mistral Large 4 |
| DTBench | 97.3% | 88.2% | GPT-6 Astra |
| LMCA | 64.4% | 35.8% | GPT-6 Astra |
| Surface Evolver Bench | Not measured | 28.1% | Only Mistral Large 4 measured |
| Furniture assembly | 80.0% | 20.8% | GPT-6 Astra |
| LMArena Text | 1441 | 1431 | GPT-6 Astra |
| LMArena Text: instruction following | 1455 | 1422 | GPT-6 Astra |
| LMArena Text: creative writing | 1420 | 1363 | GPT-6 Astra |
| LMArena Vision | 1280 | 1272 | GPT-6 Astra |
| LMArena Text: English | 1447 | 1433 | GPT-6 Astra |
| LMArena Text: non-English | 1432 | 1422 | GPT-6 Astra |
| LMArena Text: Chinese | 1489 | 1492 | Mistral Large 4 |
| LMArena Text: German | 1459 | Not measured | Only GPT-6 Astra measured |
| LMArena Text: Japanese | 1405 | 1415 | Mistral Large 4 |
| LMArena Text: Korean | 1428 | Not measured | Only GPT-6 Astra measured |
| LMArena Text: French | 1432 | 1492 | Mistral Large 4 |
| LMArena Text: Spanish | 1422 | 1455 | Mistral Large 4 |
| LMArena Text: Russian | 1438 | 1417 | GPT-6 Astra |
| LMArena Text: Polish | 1422 | Not measured | Only GPT-6 Astra measured |
| LMArena Text: multi-turn | 1447 | 1435 | GPT-6 Astra |
| LMArena Text: longer queries | 1460 | 1428 | GPT-6 Astra |
| LMArena Text: writing and language | 1437 | 1386 | GPT-6 Astra |
| LMArena Text: entertainment and media | 1427 | 1378 | GPT-6 Astra |
| LMArena Vision: English | 1280 | 1283 | Mistral Large 4 |
| LMArena Vision: Chinese | 1361 | Not measured | Only GPT-6 Astra measured |
| LMArena Vision: reading text | 1292 | 1284 | GPT-6 Astra |
| LMArena Vision: creative writing | 1286 | Not measured | Only GPT-6 Astra 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 | GPT-6 Astra | Mistral Large 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
Mistral Large 4 costs less
Repository review
50K input + 3K output tokens
Mistral Large 4 costs less
Agent loop
200K input + 10K output tokens
Mistral Large 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 | OpenAI GPT-6 Astra | Mistral Large 4 |
|---|---|---|
| Maker | OpenAI | Mistral AI |
| Weights | No public weights listed | No public weights listed |
| Released | September 4, 2026 | October 6, 2026 |
| Context window | 1.05M | 1.05M |
| Input $/Mtok | $10.00 | $1.36 |
| Output $/Mtok | $50.00 | $4.18 |
| Coding index | 85.1 | 56.4 |
| Independent measurements behind it | 5 | 5 |
| Cost per solved task | Not computable | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is GPT-6 Astra better than Mistral Large 4 for coding?
Pick GPT-6 Astra for coding: it scores 85.1 on the Vibe Coding Index against 56.4 for Mistral Large 4. Pick Mistral Large 4 for output tokens 12.0x cheaper ($4.18 vs $50.00 per million) and work measured by Finance Agent (Vals AI) or Tax Agent Bench (Vals AI), where it leads.
Which is cheaper, GPT-6 Astra or Mistral Large 4?
GPT-6 Astra costs $10.00 per million input tokens and $50.00 per million output tokens; Mistral Large 4 costs $1.36 and $4.18. Cost per solved task is known for neither of them.
Which has the bigger context window, GPT-6 Astra or Mistral Large 4?
GPT-6 Astra, with 1.05M tokens against 1.05M.
How many benchmarks is the GPT-6 Astra vs Mistral Large 4 comparison based on?
80 benchmarks have a result for both models as of October 22, 2026. GPT-6 Astra's coding index rests on 5 independent benchmarks and Mistral Large 4's on 5. Where only one model has a result, the page says so instead of naming a winner.