DeepSeek V4 Pro vs Granite 4.1 30B: Benchmarks, Pricing & Capabilities Compared
TL;DR — DeepSeek V4 Pro wins for reasoning + long-context · Granite 4.1 30B wins for general use.
DeepSeek V4 Pro DeepSeek
- Released
- 2026-04-24
- Context window
- 1M tokens
- Input price
- $1.74 / Mtok
- Output price
- $3.48 / Mtok
Granite 4.1 30B IBM
- Released
- 2026-04-29
- Context window
- 131K tokens
- Input price
- $0.00 / Mtok
- Output price
- $0.00 / Mtok
Benchmark comparison
| Benchmark | DeepSeek V4 Pro | Granite 4.1 30B |
|---|---|---|
| AA Intelligence Index | 51.5 ✓ | 14.7 |
| GPQA Diamond | 88.8% ✓ | 48.1% |
| HLE | 35.9% ✓ | 4.2% |
| IF-Bench | 76.5% ✓ | 44.4% |
| LiveCodeBench Reasoning | 66.3% ✓ | 18.7% |
| SciCode | 50.0% ✓ | 25.8% |
| TAU2-bench | 96.2% ✓ | 42.1% |
| TerminalBench-Hard | 46.2% ✓ | 2.3% |
Pricing comparison
| Metric | DeepSeek V4 Pro | Granite 4.1 30B |
|---|---|---|
| Input ($/Mtok) | $1.74 | $0.00 |
| Output ($/Mtok) | $3.48 | $0.00 |
| Cached input ($/Mtok) | — | — |
| Cost per 1M-token roundtrip (1M in + 1M out) | $5.22 | $0.00 |
Context window & modalities
| Attribute | DeepSeek V4 Pro | Granite 4.1 30B |
|---|---|---|
| Context window | 1M tokens | 131K tokens |
| Input modalities | text | text |
| Output modalities | text | text |
| Knowledge cutoff | — | — |
Verdict by use case
Coding
Insufficient data
Basis: SWE-bench
No shared coding benchmark.
Reasoning
→ DeepSeek V4 Pro
Basis: GPQA Diamond
DeepSeek V4 Pro 88.8% vs Granite 4.1 30B 48.1% on GPQA Diamond.
Math
Insufficient data
Basis: MATH / AIME
No shared math benchmark.
Long context
→ DeepSeek V4 Pro
Basis: Context window
DeepSeek V4 Pro 1M tokens vs Granite 4.1 30B 131K tokens.
Cost
→ Granite 4.1 30B
Basis: Input $/Mtok
DeepSeek V4 Pro $1.74/Mtok vs Granite 4.1 30B $0/Mtok input.
Changelog & releases
DeepSeek V4 Pro
Released 2026-04-24
Granite 4.1 30B
Released 2026-04-29