Gemini 3.8 Flash
Google · Balanced · Sep 2, 2026New
Google's newest generally available model: a fast multimodal workhorse that reads text, images, audio and video in a one-million-token context.
- Artificial Analysis Intelligence Index
- 40.9#15 of 29 · high effort
- API price per 1M tokens
- $0.75 / $3.75input / output · $1.50 blended
- Context window
- 1Mup to 66K output tokens
- Output speed
- 125 t/s24s to first answer at high effort
Benchmark results
Each row shows this model (blue) against every other model in the catalogue (grey), worst on the left, best on the right.
Overall capability
- Artificial Analysis Intelligence Index40.9#15 of 29
High effort
- Epoch Capabilities Index156.7#8 of 16
- LiveBench75.8%#15 of 23
High effort
Coding
- Terminal-Bench 4.019.1%#11 of 13
Mini-SWE-agent harness, high effort
- SWE-Bench Pro (V2 Hard)58.8%#5 of 5
Mini-swe-agent harness, high effort
- Arena WebDev1,583#17 of 27
High effort, preliminary rating
Agents & professional work
- GDPval-AA1,436#15 of 18
High effort
- Vending-Bench 2$5,094#8 of 9
Reasoning
- ARC-AGI-289.2%#5 of 11
High effort
- ARC-AGI-310.4%#3 of 5
High effort
- FrontierMath (Tiers 1–3)68.4%#11 of 14
High effort
- MathArena35.3%#10 of 10
- SimpleBench82.4%#5 of 12
Knowledge & science
- Humanity's Last Exam (Diamond)34.3%#4 of 9
- GPQA Diamond95.4%#3 of 14
High effort
Multimodal
- Arena Vision1,290#3 of 15
High effort, preliminary rating
Human preference
- Arena Text1,495#4 of 24
High effort, preliminary rating
Reasoning effort trade-offs
The same model at different thinking budgets. More effort buys a higher intelligence score at a higher cost per task and a longer wait. Measured by Artificial Analysis.
| Effort | Score | Cost / task | First answer | Speed |
|---|---|---|---|---|
| Low | 33.5 | Not measured | Not measured | Not measured |
| Medium | 39.8 | $0.93 | Not measured | Not measured |
| High (headline score) | 40.9 | $1.24 | 24s | 125 t/s |
Specs & pricing
- Developer
- Released
- Sep 2, 2026
- Availability
- Generally available
- API model ID
gemini-3.8-flash- Weights
- Proprietary (API only)
- Reasoning
- Adjustable effort
- Knowledge cutoff
- Mar 2026
- Accepts
- Text, Image, Video, Audio
- Produces
- Text
- Context window
- 1,048,576 tokens
- Max output
- 65,536 tokens
- Input price
- $0.75 / 1M tokens
- Output price
- $3.75 / 1M tokens
- Cached input
- $0.075 / 1M tokens
Introductory price until 31 December 2026; from January 2027 it doubles to $1.50 input / $7.50 output per 1M tokens.
Closest alternatives
Models from other labs nearest to Gemini 3.8 Flash on Artificial Analysis Intelligence Index.
GLM-5.3-Flash
Z.ai
Intelligence 41.8 · $0.15 / $0.50
Compare side by sideQwen3.8-Flash-Next
Alibaba
Intelligence 39.8 · $0.15 / $0.47
Compare side by sideDeepSeek-V4.1-Flash
DeepSeek
Intelligence 39.5 · $0.30 / $1.20
Compare side by side