Compare · ModelsLive · 3 picked · head to head
GPT-5.2-Codex vs MiniMax M3 vs DeepSeek V4 Pro
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
DeepSeek V4 Pro wins on 6/12 benchmarks
DeepSeek V4 Pro wins 6 of 12 shared benchmarks. Leads in math · speed · arena.
Category leads
coding·MiniMax M3reasoning·GPT-5.2-Codexlanguage·GPT-5.2-Codexmath·DeepSeek V4 Proknowledge·GPT-5.2-Codexspeed·DeepSeek V4 Proarena·DeepSeek V4 Pro
Hype vs Reality
Attention vs performance
GPT-5.2-Codex
#18 by perf·no signal
MiniMax M3
#19 by perf·no signal
DeepSeek V4 Pro
#13 by perf·no signal
Best value
DeepSeek V4 Pro
1.2x better value than MiniMax M3
GPT-5.2-Codex
9.0 pts/$
$7.88/M
MiniMax M3
93.3 pts/$
$0.75/M
DeepSeek V4 Pro
112.8 pts/$
$0.65/M
Vendor risk
Mixed exposure
One or more vendors flagged
OpenAI
$840.0B·Tier 1
MiniMax
$4.0B·Tier 1
DeepSeek
$3.4B·Tier 1
Head to head
12 benchmarks · 3 models
GPT-5.2-CodexMiniMax M3DeepSeek V4 Pro
LiveBench · Agentic Coding
MiniMax M3 leads by +3.3
GPT-5.2-Codex
51.7
MiniMax M3
60.0
DeepSeek V4 Pro
56.7
LiveBench · Coding
GPT-5.2-Codex leads by +13.6
GPT-5.2-Codex
83.6
MiniMax M3
68.2
DeepSeek V4 Pro
70.0
LiveBench · Data Analysis
GPT-5.2-Codex leads by +2.0
GPT-5.2-Codex
78.2
MiniMax M3
76.2
DeepSeek V4 Pro
74.5
LiveBench · If
GPT-5.2-Codex leads by +4.1
GPT-5.2-Codex
66.5
MiniMax M3
57.5
DeepSeek V4 Pro
62.4
LiveBench · Language
DeepSeek V4 Pro leads by +1.3
GPT-5.2-Codex
73.7
MiniMax M3
76.8
DeepSeek V4 Pro
78.1
LiveBench · Mathematics
DeepSeek V4 Pro leads by +1.9
GPT-5.2-Codex
88.8
MiniMax M3
77.0
DeepSeek V4 Pro
90.7
LiveBench · Overall
GPT-5.2-Codex leads by +0.7
GPT-5.2-Codex
74.3
MiniMax M3
70.0
DeepSeek V4 Pro
73.6
LiveBench · Reasoning
DeepSeek V4 Pro leads by +5.0
GPT-5.2-Codex
77.7
MiniMax M3
74.5
DeepSeek V4 Pro
82.7
Artificial Analysis · Agentic Index
DeepSeek V4 Pro leads by +1.0
Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?"
MiniMax M3
35.4
DeepSeek V4 Pro
36.4
Artificial Analysis · Coding Index
DeepSeek V4 Pro leads by +0.8
Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads.
MiniMax M3
58.6
DeepSeek V4 Pro
59.4
Artificial Analysis · Quality Index
MiniMax M3 leads by +0.2
MiniMax M3
44.4
DeepSeek V4 Pro
44.3
Chatbot Arena Elo · Overall
DeepSeek V4 Pro leads by +7.8
MiniMax M3
1448.2
DeepSeek V4 Pro
1455.9
Full benchmark table
| Benchmark | GPT-5.2-Codex | MiniMax M3 | DeepSeek V4 Pro |
|---|---|---|---|
LiveBench · Agentic Coding | 51.7 | 60.0 | 56.7 |
LiveBench · Coding | 83.6 | 68.2 | 70.0 |
LiveBench · Data Analysis | 78.2 | 76.2 | 74.5 |
LiveBench · If | 66.5 | 57.5 | 62.4 |
LiveBench · Language | 73.7 | 76.8 | 78.1 |
LiveBench · Mathematics | 88.8 | 77.0 | 90.7 |
LiveBench · Overall | 74.3 | 70.0 | 73.6 |
LiveBench · Reasoning | 77.7 | 74.5 | 82.7 |
Artificial Analysis · Agentic Index Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?" | — | 35.4 | 36.4 |
Artificial Analysis · Coding Index Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads. | — | 58.6 | 59.4 |
Artificial Analysis · Quality Index | — | 44.4 | 44.3 |
Chatbot Arena Elo · Overall | — | 1448.2 | 1455.9 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $1.75 | $14.00 | 400K tokens (~200 books) | $48.13 | |
| $0.30 | $1.20 | 1.0M tokens (~524 books) | $5.25 | |
| $0.43 | $0.87 | 1.0M tokens (~524 books) | $5.44 |