AI Models · Open-Source vs Open-Source

Llama 4 Maverick Open vs Mistral Large 3 2512 Open

Llama 4 Maverick vs Mistral Large 3 2512 compared — price per token, context window, multimodality, openness and which to choose. Full 2026 breakdown.

Prices & specs refreshed from live data · olud.ai

Open-model prices = cheapest provider via OpenRouter; official maker rates may be higher.

Choose Llama 4 Maverick for the lower output price ($0.8/M vs $1.5/M). It also gives you the larger 1M context window.

Llama 4 Maverick vs Mistral Large 3 2512 specs

SpecLlama 4 MaverickMistral Large 3 2512
MakerMetaMistral AI
TypeOpen-weightOpen-weight
Context window1M tokens262K tokens
Input price$0.2/M · free self-host$0.5/M · free self-host
Output price$0.8/M · free self-host$1.5/M · free self-host
Vision / multimodalYesYes
Tool / function callingYesYes
Self-hostableYesYes
LicenseLlama CommunityApache 2.0

Feature comparison

CapabilityLlama 4 MaverickMistral Large 3 2512
Open weights (downloadable)
Self-hostable
Runs fully offline
Vision / multimodal
Tool / function calling
1M+ context window

Benchmarks: Llama 4 Maverick vs Mistral Large 3 2512

Independent benchmark scores measured by Artificial Analysis. Higher is better (except latency).

Llama 4 Maverick
Mistral Large 3 2512
Intelligence index
14.3
15.9
Coding index
16.3
20.1
Math index
19.3
38
GPQA
67.1%
68%
MMLU-Pro
80.9%
80.7%
Humanity's Last Exam
4.8%
4.1%
Long Context Reasoning
46%
34.7%
LiveCodeBench
39.7%
46.5%
SciCode
33.1%
36.2%
MATH-500
88.9%
AIME
39%
AIME 2025
19.3%
38%
IFBench
43%
36.2%
τ²-Bench
17.8%
24.6%
τ-Bench Banking
3.9%
5.8%
Terminal-Bench
7.9%
12%
Terminal-Bench Hard
6.8%
15.9%
Speed
104.6 tok/s
60.9 tok/s
Latency
0.66s
1.05s
Intelligence per $
34.5
21.2

Benchmark data by Artificial Analysis.

How Llama 4 Maverick and Mistral Large 3 2512 score

🏆 Best value & openness: Llama 4 Maverick (5.0 vs 4.7 / 5)
CriterionLlama 4 MaverickMistral Large 3 2512
Cost-efficiency5.04.5
Context window5.04.0
Openness5.05.0
Self-hosting5.05.0
Multimodality5.05.0

Scores come from live data — output price (cost), context length, open vs closed weights (openness & self-hosting) and vision/tool support (multimodality). Raw task quality isn't scored here; it depends on your benchmark — see the verdict.

What each model is

Llama 4 Maverick Open

Meta · Open-weight

Llama 4 Maverick 17B Instruct (128E) is a high-capacity multimodal language model from Meta, built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters per forward...

Mistral Large 3 2512 Open

Mistral AI · Open-weight

Mistral Large 3 2512 is Mistral’s most capable model to date, featuring a sparse mixture-of-experts architecture with 41B active parameters (675B total), and released under the Apache 2.0 license.

Other models in these families

These variants are tracked but not compared here — one page per family keeps the comparison readable.

Other variants tracked
Llama 4 ScoutR1 Distill Llama 70BLlama 3.3 70B InstructLlama 3.1 8B InstructLlama 3.1 70B InstructHermes 3 70B InstructLlama 3.2 3B InstructLlama 3.2 1B InstructAion-RP 1.0 (8B)Hermes 3 405B InstructLlama 3.1 Euryale 70B v2.2Llama 3.3 Euryale 70BLlama Guard 4 12BLlama 3 8B Lunaris
Other variants tracked
Mistral Medium 3.5Mistral Small 4Devstral 2 2512Mistral Medium 3.1Mistral Small 3.1 24BMistral Medium 3Ministral 3 14B 2512Mistral Small 3.2 24BMinistral 3 8B 2512Mistral LargeMistral Large 2407Mistral Small 3SabaMinistral 3 3B 2512Mixtral 8x22B InstructCodestral 2508UncensoredVoxtral Small 24B 2507Mistral Nemo

Frequently asked questions

Llama 4 Maverick vs Mistral Large 3 2512 — which is cheaper?

Llama 4 Maverick is cheaper on output ($0.8/M vs $1.5/M).

Which has the larger context window?

Llama 4 Maverick offers the larger context window (1M tokens).

Llama 4 Maverick vs Mistral Large 3 2512 — which should I pick in 2026?

Choose Llama 4 Maverick for the lower output price ($0.8/M vs $1.5/M). It also gives you the larger 1M context window.

People also compare

Explore more open-source AI

Browse the full open-source model leaderboard, thousands of tools and live benchmarks — all in one place.

Open the leaderboard →