vs
1:10
overall capability wins (0 ties across 11 scored). DeepSeek V4 Flash 0731 leads overall.
- Reasoning0:5
- Coding0:2
- Tool use0:1
- Composite indices0:2
- Pricing0:3
- Specs1:2
Verdict
DeepSeek V4 Flash 0731 leads coding (widest gap: +59.6 on TermBench); DeepSeek V4 Flash 0731 offers ~7.8x larger context (1M vs 128K); DeepSeek V4 Flash 0731 allows ~6.0x more max output (384K vs 64K).
- DeepSeek V4 Flash 0731 leads coding (widest gap: +59.6 on TermBench)
- DeepSeek V4 Flash 0731 offers ~7.8x larger context (1M vs 128K)
- DeepSeek V4 Flash 0731 allows ~6.0x more max output (384K vs 64K)
- DeepSeek V4 Flash 0731 is ~1.5x cheaper on a blended token basis than DeepSeek R1
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- DeepSeek R1
- $1.10
- DeepSeek V4 Flash 0731
- $0.77
DeepSeek V4 Flash 0731 is about 1.4x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
DeepSeek
DeepSeek
Family
DeepSeek
DeepSeek
License
MIT
MIT
Open weights
Yes
Yes
Release
Jan 20, 2025
Jul 31, 2026
Knowledge cutoff
-
-
API / provider
DeepSeek
DeepSeek
Modalities
text → text
text → text
Specs
Context window
128K
1M
Max output
64K
384K
Parameters
671B (37B act.)
304B
Pricing
Input $/1M
$0.55
$0.44
Output $/1M
$2.19
$1.32
Blended $/1M (3∶1)
$0.96
$0.66
Speed
tok/s
45
-
TTFT (s)
1.2
-
Reasoning
MMLU-Pro
84.0
—
GPQA Diamond
70.8
90.8
Humanity's Last Exam
8.5
38.6
AIME 2025
70.0
—
MATH-500
97.3
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
30.5
40.4
AA-LCR v1.1
57.7
79.7
CritPt
0.6
16.6
MMMU-Pro
—
—
IFBench
39.0
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
49.2
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
19.1
78.7
Aider Polyglot
56.9
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
38.3
50.3
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
54.2
DeepSWE
—
54.4
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,398
—
| Metric | DeepSeek R1 | DeepSeek V4 Flash 0731 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | DeepSeek | DeepSeek | — |
| Family | DeepSeek | DeepSeek | — |
| License | MIT | MIT | — |
| Open weights | Yes | Yes | — |
| Release | Jan 20, 2025 | Jul 31, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | DeepSeek | DeepSeek | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 128K | 1M | B +872K |
| Max output | 64K | 384K | B +320K |
| Parameters | 671B (37B act.) | 304B | A +367 |
| Pricing | |||
| Input $/1M | $0.55 | $0.44 | B +$0.11 |
| Output $/1M | $2.19 | $1.32 | B +$0.87 |
| Blended $/1M (3∶1) | $0.96 | $0.66 | B +$0.30 |
| Speed | |||
| tok/s | 45 | - | — |
| TTFT (s) | 1.2 | - | — |
| Reasoning | |||
| MMLU-Pro | 84.0 | — | — |
| GPQA Diamond | 70.8 | 90.8 | B +20.0 pts |
| Humanity's Last Exam | 8.5 | 38.6 | B +30.1 pts |
| AIME 2025 | 70.0 | — | — |
| MATH-500 | 97.3 | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 30.5 | 40.4 | B +9.9 pts |
| AA-LCR v1.1 | 57.7 | 79.7 | B +22.0 pts |
| CritPt | 0.6 | 16.6 | B +16.0 pts |
| MMMU-Pro | — | — | — |
| IFBench | 39.0 | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | 49.2 | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | 19.1 | 78.7 | B +59.6 pts |
| Aider Polyglot | 56.9 | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 38.3 | 50.3 | B +12.0 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | 54.2 | — |
| DeepSWE | — | 54.4 | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,398 | — | — |
DeepSeek R1
- Context
- 128K / 64K out
- Parameters
- 671B (37B act.)
- Price
- $0.55 / $2.19
- Speed
- 45 tok/s · 1.2s TTFT
- Modalities
- text → text
- License
- MIT
Original R1 reasoning model - strong math with explicit chain-of-thought; superseded by R1-0528 / V4 for coding.
DeepSeek V4 Flash 0731
- Context
- 1M / 384K out
- Parameters
- 304B
- Price
- $0.44 / $1.32 peak · $0.22 / $0.66 off-peak
- Speed
- —
- Modalities
- text → text
- License
- MIT
Official DeepSeek-V4-Flash release (Jul 31, 2026 public beta) superseding the preview - same base architecture plus a DSpark speculative-decoding module (304B total), MIT weights on Hugging Face, and agent benchmarks well above V4-Pro-Preview (Terminal-Bench 2.1 82.7, NL2Repo 54.2). Peak/off-peak API pricing took effect Aug 16, 2026: $0.44 in / $1.32 out peak, $0.22 in / $0.66 out off-peak (per 1M tokens).
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.