DeepSeek V3.2 vs Llama 4 Maverick

Put models side by side on price, context, capabilities, and benchmarks.

DeepSeek V3.2 is DeepSeek's V3-generation model upgrade underlying both the deepseek-chat (non-thinking) and deepseek-reasoner (thinking) modes, improving efficiency and reasoning quality over V3.1. It is designed as a strong, low-cost general-purpose and reasoning model.

Mathematical#96Writing, Literature, & Language#108Life, Physical, & Social Science#110Entertainment, Sports, & Media#119
Learn more
AuthorDeepSeek
CountryπŸ‡¨πŸ‡³ China
ReleasedDec 1, 2025
Overview
Cost rate
0.1x
Context
164K
Input
Output
Reasoning
Tool calling
Structured outputs
Pricing/M tokens
Input$0.26
Cached input$0.13
Output$0.38
Context
Context length163,840
Max output tokens147,456
Knowledge cutoff-
Performance
Throughput-
Latency-

Llama 4 Maverick is Meta's high-capability open-weight Mixture-of-Experts model, offering strong multimodal (text and image) understanding, reasoning, and coding at efficient inference cost. It targets general-purpose assistant and agentic use cases needing frontier-adjacent quality in an open model.

Image understanding#122Legal & Government#242Medicine & Healthcare#242Writing, Literature, & Language#243
Learn more
AuthorMeta
CountryπŸ‡ΊπŸ‡Έ United States
ReleasedApr 5, 2025
Overview
Cost rate
0.2x
Context
524K
Input
Output
Reasoning
Tool calling
Structured outputs
Pricing/M tokens
Input$0.35
Cached input$0.17
Output$1.00
Context
Context length524,288
Max output tokens32,768
Knowledge cutoff2024-08-31
Performance
Throughput81 tok/s
Latency1.03s

Benchmarks

Head-to-head scores across reasoning, coding, and agentic indices, plus per-domain rankings.

Intelligence

64.6
Claude Fable 5.2 (Max, Default Fallback)
60.0
Gemini 4 Carbon (High)
59.1
Claude Opus 5.5 (Max, Default Fallback)
56.9
GPT-6 Astra (Max)
54.6
GPT-6.1 Sol (Max)
54.5
Claude Sonnet 5.5 (Max, Default Fallback)

DeepSeek V3.2 has no data

Llama 4 Maverick has no data

Coding

80.7
Claude Fable 5.1
78.3
GPT-5.6 Sol
78.0
Claude Opus 5
76.9
GPT-6 Astra
44.2
DeepSeek V3.2
16.3
Llama 4 Maverick

Automation & Tool Use

57.2
Claude Fable 5.1
56.5
Claude Opus 5
56.0
Qwen3.8 Max (0902)
55.5
Muse Spark 1.3
53.1
GLM 5.3
0.6
Llama 4 Maverick

DeepSeek V3.2 has no data

Intelligence, coding & agentic scores from Artificial Analysis.

Software & IT Services

arena.ai
1549
Gemini 4 Carbon (High)
1548
Gemini 4 Argon (High)
1540
Claude Opus 4.7 (High)
1540
Muse Wonder (Max)
1538
Claude Fable 5 (High)
1537
Claude Opus 4.6 (High)
1534
Claude Opus 5.5 (High)
1532
Muse Spark 1.2 (xHigh)
1464
DeepSeek V3.2
1358
Llama 4 Maverick

Popular comparisons