vs
0:9
overall capability wins (0 ties across 9 scored). DeepSeek V4 Flash leads overall.
- Reasoning0:5
- Coding0:1
- Pricing0:3
- Speed0:2
- Specs0:3
Verdict
DeepSeek V4 Flash leads coding (widest gap: +14.0 on SciCode); DeepSeek V4 Flash offers ~7.8x larger context (1M vs 128K); DeepSeek V4 Flash allows ~46.9x more max output (384K vs 8K).
- DeepSeek V4 Flash leads coding (widest gap: +14.0 on SciCode)
- DeepSeek V4 Flash offers ~7.8x larger context (1M vs 128K)
- DeepSeek V4 Flash allows ~46.9x more max output (384K vs 8K)
- DeepSeek V4 Flash is ~4.6x cheaper on a blended token basis than DeepSeek-R1-Distill-Llama-70B
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- DeepSeek-R1-Distill-Llama-70B
- $1
- DeepSeek V4 Flash
- $0.21
DeepSeek V4 Flash is about 4.8x 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
Apr 24, 2026
Knowledge cutoff
-
-
API / provider
Together / Fireworks (ref.)
DeepSeek
Modalities
text → text
text → text
Specs
Context window
128K
1M
Max output
8K
384K
Parameters
70B
284B (13B act.)
Pricing
Input $/1M
$0.80
$0.14
Output $/1M
$0.80
$0.28
Blended $/1M (3∶1)
$0.80
$0.18
Speed
tok/s
55
120
TTFT (s)
0.45
0.25
Reasoning
MMLU-Pro
79.5
—
GPQA Diamond
40.2
89.4
Humanity's Last Exam
5.1
34.8
AIME 2025
53.7
—
MATH-500
94.5
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
19.2
36.8
AA-LCR v1.1
10.0
74.3
CritPt
—
7.1
MMMU-Pro
—
—
IFBench
27.6
79.2
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
57.5
—
Terminal-Bench 2.1
—
61.8
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
35.3
—
SciCode
31.3
45.3
CursorBench
—
—
SWE-Rebench
—
38.4
NL2Repo-Bench
—
39.4
DeepSWE
—
7.3
WebDev Arena
—
1,580
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
1,436
| Metric | DeepSeek-R1-Distill-Llama-70B | DeepSeek V4 Flash | Delta |
|---|---|---|---|
| Identity | |||
| Organization | DeepSeek | DeepSeek | — |
| Family | DeepSeek | DeepSeek | — |
| License | MIT | MIT | — |
| Open weights | Yes | Yes | — |
| Release | Jan 20, 2025 | Apr 24, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Together / Fireworks (ref.) | DeepSeek | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 128K | 1M | B +872K |
| Max output | 8K | 384K | B +376K |
| Parameters | 70B | 284B (13B act.) | B +214 |
| Pricing | |||
| Input $/1M | $0.80 | $0.14 | B +$0.66 |
| Output $/1M | $0.80 | $0.28 | B +$0.52 |
| Blended $/1M (3∶1) | $0.80 | $0.18 | B +$0.63 |
| Speed | |||
| tok/s | 55 | 120 | B +65 |
| TTFT (s) | 0.45 | 0.25 | B +0.20 |
| Reasoning | |||
| MMLU-Pro | 79.5 | — | — |
| GPQA Diamond | 40.2 | 89.4 | B +49.2 pts |
| Humanity's Last Exam | 5.1 | 34.8 | B +29.7 pts |
| AIME 2025 | 53.7 | — | — |
| MATH-500 | 94.5 | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 19.2 | 36.8 | B +17.6 pts |
| AA-LCR v1.1 | 10.0 | 74.3 | B +64.3 pts |
| CritPt | — | 7.1 | — |
| MMMU-Pro | — | — | — |
| IFBench | 27.6 | 79.2 | B +51.6 pts |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | 57.5 | — | — |
| Terminal-Bench 2.1 | — | 61.8 | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | 35.3 | — | — |
| SciCode | 31.3 | 45.3 | B +14.0 pts |
| CursorBench | — | — | — |
| SWE-Rebench | — | 38.4 | — |
| NL2Repo-Bench | — | 39.4 | — |
| DeepSWE | — | 7.3 | — |
| WebDev Arena | — | 1,580 | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | 1,436 | — |
DeepSeek-R1-Distill-Llama-70B
- Context
- 128K / 8K out
- Parameters
- 70B
- Price
- $0.80 / $0.80
- Speed
- 55 tok/s · 0.45s TTFT
- Modalities
- text → text
- License
- MIT
Llama-70B distilled from R1 reasoning traces — open reasoning without full R1 compute cost.
DeepSeek V4 Flash
- Context
- 1M / 384K out
- Parameters
- 284B (13B act.)
- Price
- $0.14 / $0.28
- Speed
- 120 tok/s · 0.25s TTFT
- Modalities
- text → text
- License
- MIT
Budget DeepSeek V4 API tier (Apr 2026 preview). Superseded by DeepSeek-V4-Flash-0731 (public beta), which re-post-trains the same architecture and reports far stronger agent benchmarks.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.