# GPT-5.4 vs Mixtral 8x22B

> GPT-5.4 is the stronger model overall, scoring 59.4 to 27.1 on the Noometry Index. Mixtral 8x22B costs 1.9× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gpt-5-4-vs-mixtral-8x22b
- Last updated: 2026-10-10
- Shared benchmarks: 22

## Summary

- They share 22 benchmarks with published results for both. GPT-5.4 scores higher in 9 categories and Mixtral 8x22B in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.4 leads 73.5 to 22.9.
- The biggest single-benchmark swing is WeirdML: 77.7% for GPT-5.4 and 3.2% for Mixtral 8x22B.
- Mixtral 8x22B is cheaper at $2 / $6 per million input/output tokens, against $2.50 / $15 for GPT-5.4.
- GPT-5.4 accepts more context: 1.05M tokens versus 64K.
- Mixtral 8x22B has downloadable open weights; the other is API-only.

## Snapshot

| | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| Provider | OpenAI | Mistral AI |
| Noometry Index | 59.4 | 27.1 |
| Rank | 16 | 333 |
| Context | 1.05M | 64K |
| Input $/M | $2.50 | $2 |
| Output $/M | $15 | $6 |
| Weights | Proprietary | Open |

## Coding

- GPT-5.4: 52.6 (#33)
- Mixtral 8x22B: 24.2 (#329)

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| WeirdML | 77.7% | 3.2% |
| LMArena Coding | 1497 | 1166 |
| SWE-bench Verified | 76.9% | — |
| DeepSWE | 51.8% | — |
| LMArena WebDev | 1465 | — |
| SciCode | 56.6% | — |
| GSO | 31.4% | — |
| BigCodeBench Instruct | — | 40.6% |
| MirrorCode | 15.6% | — |
| BigCodeBench Complete | — | 50.2% |
| ALE-Bench | 1,607 | — |
| AlgoTune | 1.85 | — |
| HumanEval+ | — | 72% |
| MBPP+ | — | 64.3% |

## Agentic & Tool Use

- GPT-5.4: 46.5 (#13)
- Mixtral 8x22B: 23.1 (#127)

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| Terminal-Bench | 81.8% | — |
| APEX-Agents | 52.4% | — |
| τ²-bench Banking | 39.4% | — |
| Cybench | — | 7.5% |
| DeepResearch Bench | 35.1% | — |
| PostTrainBench | 19% | — |
| GBAEval | 45.1% | — |
| LMArena Search | 1197 | — |
| METR Time Horizons | 74.3% | — |
| Vending-Bench 2 | 6,144 | — |

## Reasoning

- GPT-5.4: 61.8 (#19)
- Mixtral 8x22B: 19.9 (#248)

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| LMArena Hard Prompts | 1485 | 1150 |
| DTBench | 94.4% | 55.1% |
| Epoch Capabilities Index | 156.81 | 122.03 |
| ForecastBench | 59.5 | 56.3 |
| ARC-AGI-2 | 74% | — |
| Kagi LLM Benchmark | 63.8% | — |
| NYT Connections (extended) | 91.3% | — |
| ARC-AGI-1 | 93.7% | — |
| CritPt | 23.4% | — |
| Chess Puzzles | 44% | — |
| EnigmaEval | 16% | — |
| Thematic Generalization | 80% | — |
| EBR-Bench | 25.4% | — |
| Mystery Game Puzzles | 37% | — |
| LMCA | 52% | — |

## Math

- GPT-5.4: 73.5 (#19)
- Mixtral 8x22B: 22.9 (#275)

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| LMArena Math | 1488 | 1184 |
| FrontierMath (Tiers 1-3) | 78.6% | — |
| FrontierMath Tier 4 | 49% | — |
| MathArena Final-Answer Competitions | 83.1% | — |
| OTIS Mock AIME 2024-2025 | 97.8% | — |
| ProofBench | 56% | — |
| Omni-MATH | — | 16.3% |
| MATH Level 5 | — | 24.2% |
| FrontierMath (Feb 2025 set) | 47.6% | — |
| FrontierMath Tier 4 (v1) | 27.1% | — |

## Knowledge

- GPT-5.4: 65.3 (#14)
- Mixtral 8x22B: 15.1 (#293)

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| GPQA Diamond | 93.3% | 34.1% |
| LMArena Expert | 1507 | 1113 |
| Humanity's Last Exam | 36.2% | — |
| SimpleQA Verified | 45.1% | — |
| MMLU-Pro | — | 46% |
| Vectara Hallucination Rate | 7% | — |
| GPQA (HELM) | — | 33.4% |
| MMLU | — | 77.8% |

## Multimodal

- GPT-5.4: 43.7 (#20)
- Mixtral 8x22B: —

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| LMArena Vision | 1303 | — |
| Blueprint-Bench 2 | 27.1% | — |
| Furniture Assembly | 37.5% | — |
| LMArena Document | 1471 | — |

## Multilingual

- GPT-5.4: 56.2 (#23)
- Mixtral 8x22B: 32.8 (#255)

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| LMArena Non-English | 1465 | 1128 |
| LMArena Chinese | 1519 | 1116 |
| LMArena French | 1493 | 1166 |
| LMArena German | 1472 | 1141 |
| LMArena Japanese | 1485 | 1037 |
| LMArena Korean | 1448 | 1057 |
| LMArena Russian | 1480 | 1158 |
| LMArena Spanish | 1454 | 1151 |

## Instruction Following

- GPT-5.4: 77.1 (#27)
- Mixtral 8x22B: 57.7 (#266)

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| LMArena Instruction Following | 1469 | 1147 |
| IFEval | — | 72.4% |

## Long Context

- GPT-5.4: 50.3 (#8)
- Mixtral 8x22B: 34.7 (#247)

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| LMArena Longer Query | 1473 | 1144 |
| CL-bench | 27.9% | — |
| CL-bench Life | 21.7% | — |

## Writing & Preference

- GPT-5.4: 71.9 (#17)
- Mixtral 8x22B: 36.9 (#262)

| Benchmark | GPT-5.4 | Mixtral 8x22B |
|---|---|---|
| LMArena Text | 1469 | 1162 |
| LMArena Creative Writing | 1439 | 1141 |
| LMArena Multi-Turn | 1482 | 1130 |
| EQ-Bench Creative Writing | 1840 | — |
| WildBench | — | 71.1% |
| EQ-Bench 4 | 1272 | — |

## FAQ

### Is GPT-5.4 better than Mixtral 8x22B?

GPT-5.4 is the stronger model overall, scoring 59.4 to 27.1 on the Noometry Index. Mixtral 8x22B costs 1.9× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.

### Which is cheaper, GPT-5.4 or Mixtral 8x22B?

Mixtral 8x22B is cheaper. It lists at $2 per million input tokens and $6 per million output tokens; GPT-5.4 lists at $2.50 and $15.

### Is GPT-5.4 or Mixtral 8x22B better for coding?

GPT-5.4 scores higher on coding benchmarks: 52.6 versus 24.2 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.4 does, with 1.05M tokens against 64K.

### How many benchmarks do GPT-5.4 and Mixtral 8x22B share?

22 benchmarks have published results for both models. GPT-5.4 has 68 scored results on Noometry and Mixtral 8x22B has 34.