vs
1:1
overall capability wins (0 ties across 2 scored). Price/speed category wins do not decide overall. Tied overall.
- Specs1:1
Verdict
Gemini 1.5 Pro offers ~1.9x larger context (2M vs 1.0M); Gemini 3.8 Flash Cyber allows ~8.0x more max output (66K vs 8K).
- Gemini 1.5 Pro offers ~1.9x larger context (2M vs 1.0M)
- Gemini 3.8 Flash Cyber allows ~8.0x more max output (66K vs 8K)
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- Gemini 1.5 Pro
- $2.50
- Gemini 3.8 Flash Cyber
- -
Price per 1M tokens
Full comparison
Organization
Family
Gemini
Gemini
License
Proprietary
Proprietary
Open weights
No
No
Release
Feb 15, 2024
Sep 2, 2026
Knowledge cutoff
-
-
API / provider
No primary API price listed
Modalities
text, image, audio, video → text
text → text
Specs
Context window
2M
1.0M
Max output
8K
66K
Parameters
—
—
Pricing
Input $/1M
$1.25
—
Output $/1M
$5
—
Blended $/1M (3∶1)
$2.19
-
Speed
tok/s
70
-
TTFT (s)
0.5
-
Reasoning
MMLU-Pro
—
—
GPQA Diamond
58.9
—
Humanity's Last Exam
4.6
—
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
—
—
AA-LCR v1.1
—
—
CritPt
—
—
MMMU-Pro
55.0
—
IFBench
—
—
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
—
—
SWE-bench Pro
—
—
SWE-bench Multilingual
—
—
LiveCodeBench
—
—
Terminal-Bench 2.1
—
—
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
29.5
—
CursorBench
—
—
SWE-Rebench
—
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
—
—
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
—
—
| Metric | Gemini 1.5 Pro | Gemini 3.8 Flash Cyber | Delta |
|---|---|---|---|
| Identity | |||
| Organization | — | ||
| Family | Gemini | Gemini | — |
| License | Proprietary | Proprietary | — |
| Open weights | No | No | — |
| Release | Feb 15, 2024 | Sep 2, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | No primary API price listed | — | |
| Modalities | text, image, audio, video → text | text → text | — |
| Specs | |||
| Context window | 2M | 1.0M | A +951K |
| Max output | 8K | 66K | B +57K |
| Parameters | — | — | — |
| Pricing | |||
| Input $/1M | $1.25 | — | — |
| Output $/1M | $5 | — | — |
| Blended $/1M (3∶1) | $2.19 | - | — |
| Speed | |||
| tok/s | 70 | - | — |
| TTFT (s) | 0.5 | - | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 58.9 | — | — |
| Humanity's Last Exam | 4.6 | — | — |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | — | — | — |
| AA-LCR v1.1 | — | — | — |
| CritPt | — | — | — |
| MMMU-Pro | 55.0 | — | — |
| IFBench | — | — | — |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | — | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | — | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | — | — | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 29.5 | — | — |
| CursorBench | — | — | — |
| SWE-Rebench | — | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | — | — | — |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | — | — | — |
Gemini 1.5 Pro
- Context
- 2M / 8K out
- Parameters
- —
- Price
- $1.25 / $5
- Speed
- 70 tok/s · 0.5s TTFT
- Modalities
- text, image, audio, video → text
- License
- Proprietary
Landmark 1–2M token multimodal model that popularized long-video and codebase-scale context.
Gemini 3.8 Flash Cyber
- Context
- 1.0M / 66K out
- Parameters
- —
- Price
- —
- Speed
- —
- Modalities
- text → text
- License
- Proprietary
Cybersecurity-specialized Gemini 3.8 Flash fine-tune for autonomous vulnerability discovery and automated patching - limited to the Fairwind Program for governments, critical-infrastructure operators, and trusted software maintainers (no public API).
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.
These two models have no overlapping published benchmarks in our dataset. Compare specs and pricing instead.