Claude Opus 5.5
Muse Wonder
MiMo-V2.6-Pro
Qwen3.8-Max
Kimi K3
GLM-5.3
GPT-5.5
ERNIE 5.1
DeepSeek-V4.1-Flash-20260910
Grok 4.20 Beta
amazon-nova-experimental-chat
Dola Seed 2.0 Pro
NVIDIA-Nemotron-3-Ultra-550B-A55B
Gemma 4 31B
Step 5 Preview
Gemini 3.8 Flash
Gemini 3.8 Flash is a proprietary model from Google, released Sep 2, 2026. It ranks #8 of 351 on the LMArena text leaderboard with a rating of 1498, and #15 of 27 on the Vals Index with 55.95%. It costs $0.82 per million input tokens and $4.13 per million output tokens, with a 1M-token context window.
- Developer
- Google DeepMind
- Released
- Sep 2, 2026
- Context window
- 1M
- Price in / out, per 1M tokens
- $0.82 / $4.13
- Weights
- Proprietary
- Inputs
- Text, Images, Video, Files, Audio
- Reasoning
- Yes
- Tool use
- Yes
Benchmarks
| Benchmark | Score bar | Accuracy | Rank | Cost / test |
|---|---|---|---|---|
| Vals IndexVals AI | 55.95% | 15 / 27 | $6.06 | |
| Terminal-Bench 4.0Coding | 19.70% | 31 / 55 | $8.65 | |
| SWE-benchCoding | 80.00% | 23 / 85 | $2.19 | |
| Vibe Code BenchCoding | 78.65% | 29 / 115 | $6.87 | |
| Code MigrationCoding | 38.65% | 22 / 28 | $15.79 | |
| LiveCodeBenchCoding | 89.48% | 3 / 143 | $1.5 / $7.5 | |
| Finance AgentFinance | 65.76% | 5 / 31 | $1.99 | |
| Finance IndexFinance | 60.46% | 9 / 31 | $3.70 | |
| Tax Agent BenchFinance | 32.12% | 21 / 76 | $0.86 | |
| Legal IndexLaw | 69.78% | 14 / 28 | $2.05 | |
| Legal Research BenchLaw | 38.94% | 34 / 85 | $1.63 | |
| MedCodeMedicine | 48.13% | 31 / 107 | $1.5 / $7.5 | |
| GPQA DiamondReasoning | 94.44% | 4 / 138 | $1.5 / $7.5 | |
| MMLU-ProReasoning | 90.22% | 5 / 138 | $1.5 / $7.5 | |
| MMMUReasoning | 89.08% | 4 / 93 | $1.5 / $7.5 | |
| Terminal-Bench ScienceReasoning | 8.57% | 19 / 42 | $5.64 |
Vals AI, as of Oct 24, 2026.
LMArena, by category
| Leaderboard | Rating | Rank | Votes |
|---|---|---|---|
| TextLMArena | 1498 | 8 / 351 | 44K |
| CodingLMArena | 1508 | 17 / 346 | 11.4K |
| MathLMArena | 1532 | 4 / 339 | 2.3K |
| Creative writingLMArena | 1489 | 8 / 349 | 9.8K |
| Instruction followingLMArena | 1492 | 11 / 351 | 15.8K |
| Hard promptsLMArena | 1509 | 9 / 351 | 28.3K |
| Multi-turnLMArena | 1503 | 9 / 349 | 6.6K |
| Long queriesLMArena | 1507 | 9 / 330 | 20.2K |
| ExpertLMArena | 1527 | 11 / 305 | 5.1K |
| Legal and governmentIndustries | 1498 | 13 / 325 | 3.9K |
| Medicine and healthcareIndustries | 1506 | 7 / 325 | 3.1K |
| Business and financeIndustries | 1479 | 14 / 344 | 7.9K |
| ChineseLanguages | 1553 | 7 / 328 | 3.6K |
| SpanishLanguages | 1485 | 13 / 249 | 1.5K |
| GermanLanguages | 1509 | 5 / 261 | 684 |
| FrenchLanguages | 1495 | 17 / 252 | 1.5K |
| JapaneseLanguages | 1511 | 6 / 236 | 794 |
| WebDevLMArena | 1585 | 36 / 124 | 11.4K |
| VisionLMArena | 1315 | 7 / 146 | 4.6K |
LMArena, CC BY 4.0, as of Oct 22, 2026. Where a model is rated at several effort levels, its best is shown.
Compare Gemini 3.8 Flash
Google DeepMind
Google’s AI lab, behind Gemini and the open Gemma family, with very long context and native multimodality. Founded 2010 (DeepMind), merged with Google Brain in 2023, based in London. deepmind.google
Questions
Who makes Gemini 3.8 Flash?
Gemini 3.8 Flash is made by Google DeepMind, founded 2010 (DeepMind), merged with Google Brain in 2023 and based in London.
How much does Gemini 3.8 Flash cost?
$0.82 per million input tokens and $4.13 per million output tokens, at its provider’s list price.
What is Gemini 3.8 Flash’s context window?
1M tokens (1,048,576).
How does Gemini 3.8 Flash rank?
#8 of 351 on LMArena’s text leaderboard, rated 1498 (95% interval 1494–1503) from 43,997 votes, as of Oct 22, 2026.
Is Gemini 3.8 Flash open source?
No. Gemini 3.8 Flash is proprietary and served through its provider’s API.