vs
1:8
overall capability wins (0 ties across 9 scored). DeepSeek V4.1 Flash leads overall.
- Reasoning0:4
- Coding0:2
- Pricing3:0
- Specs1:2
Verdict
DeepSeek V4.1 Flash leads coding (widest gap: +295.1 on WebDev Elo); DeepSeek V4.1 Flash offers ~6.1x larger context (1M vs 164K); DeepSeek V4.1 Flash allows ~6.0x more max output (384K vs 64K).
- DeepSeek V4.1 Flash leads coding (widest gap: +295.1 on WebDev Elo)
- DeepSeek V4.1 Flash offers ~6.1x larger context (1M vs 164K)
- DeepSeek V4.1 Flash allows ~6.0x more max output (384K vs 64K)
- DeepSeek V3.2 is ~1.7x cheaper on a blended token basis than DeepSeek V4.1 Flash
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- DeepSeek V3.2
- $0.39
- DeepSeek V4.1 Flash
- $0.60
DeepSeek V3.2 is about 1.6x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
DeepSeek
DeepSeek
Family
DeepSeek
DeepSeek
License
MIT
MIT
Open weights
Yes
Yes
Release
Nov 20, 2025
Sep 10, 2026
Knowledge cutoff
-
-
API / provider
DeepSeek
DeepSeek
Modalities
text → text
text, image → text
Specs
Context window
164K
1M
Max output
64K
384K
Parameters
685B (37B act.)
552B (16B act.)
Pricing
Input $/1M
$0.28
$0.30
Output $/1M
$0.42
$1.20
Blended $/1M (3∶1)
$0.32
$0.52
Speed
tok/s
90
-
TTFT (s)
0.35
-
Reasoning
MMLU-Pro
86.2
—
GPQA Diamond
84.0
—
Humanity's Last Exam
24.6
39.2
AIME 2025
92.0
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
33.0
46.4
AA-LCR v1.1
73.3
84.0
CritPt
2.9
14.3
MMMU-Pro
—
77.0
IFBench
60.7
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
70.0
—
SWE-bench Pro
15.6
—
SWE-bench Multilingual
59.0
—
LiveCodeBench
86.2
—
Terminal-Bench 2.1
46.8
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
38.9
51.9
CursorBench
—
—
SWE-Rebench
48.9
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
1,325
1,620
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,425
—
| Metric | DeepSeek V3.2 | DeepSeek V4.1 Flash | Delta |
|---|---|---|---|
| Identity | |||
| Organization | DeepSeek | DeepSeek | — |
| Family | DeepSeek | DeepSeek | — |
| License | MIT | MIT | — |
| Open weights | Yes | Yes | — |
| Release | Nov 20, 2025 | Sep 10, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | DeepSeek | DeepSeek | — |
| Modalities | text → text | text, image → text | — |
| Specs | |||
| Context window | 164K | 1M | B +836K |
| Max output | 64K | 384K | B +320K |
| Parameters | 685B (37B act.) | 552B (16B act.) | A +133 |
| Pricing | |||
| Input $/1M | $0.28 | $0.30 | A +$0.02 |
| Output $/1M | $0.42 | $1.20 | A +$0.78 |
| Blended $/1M (3∶1) | $0.32 | $0.52 | A +$0.21 |
| Speed | |||
| tok/s | 90 | - | — |
| TTFT (s) | 0.35 | - | — |
| Reasoning | |||
| MMLU-Pro | 86.2 | — | — |
| GPQA Diamond | 84.0 | — | — |
| Humanity's Last Exam | 24.6 | 39.2 | B +14.6 pts |
| AIME 2025 | 92.0 | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 33.0 | 46.4 | B +13.4 pts |
| AA-LCR v1.1 | 73.3 | 84.0 | B +10.7 pts |
| CritPt | 2.9 | 14.3 | B +11.4 pts |
| MMMU-Pro | — | 77.0 | — |
| IFBench | 60.7 | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | 70.0 | — | — |
| SWE-bench Pro | 15.6 | — | — |
| SWE-bench Multilingual | 59.0 | — | — |
| LiveCodeBench | 86.2 | — | — |
| Terminal-Bench 2.1 | 46.8 | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 38.9 | 51.9 | B +13.0 pts |
| CursorBench | — | — | — |
| SWE-Rebench | 48.9 | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | 1,325 | 1,620 | B +295 Elo |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,425 | — | — |
DeepSeek V3.2
- Context
- 164K / 64K out
- Parameters
- 685B (37B act.)
- Price
- $0.28 / $0.42
- Speed
- 90 tok/s · 0.35s TTFT
- Modalities
- text → text
- License
- MIT
Prior DeepSeek V3.x generation - still excellent open-weight coding value.
DeepSeek V4.1 Flash
- Context
- 1M / 384K out
- Parameters
- 552B (16B act.)
- Price
- $0.30 / $1.20 peak · $0.15 / $0.60 off-peak
- Speed
- —
- Modalities
- text, image → text
- License
- MIT
New default DeepSeek Flash tier (Sep 10, 2026), the smallest model in a new Causal Encoder-Decoder architecture family: 552B total parameters activating 8B (prefill) / 16B (decode) per token, MIT-licensed weights on Hugging Face. Replaces DeepSeek-V4-Flash and V4-Flash-Vision-Exp, which are retired but still routed to this model for compatibility. DeepSeek's own reported numbers (GPQA Diamond 90.9, Terminal-Bench 2.1 90.6, text-only HLE 39.1) are self-measured and not directly comparable to this catalog's Artificial-Analysis-sourced cells. API pricing: $0.30 in / $1.20 out peak, $0.15 in / $0.60 out off-peak (per 1M tokens); off-peak cache-hit input is $0.003.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.