Gemma 2 9B It
Jun 2024
Gemma 2 9B ItCompare inon
Benchmarks of Gemma 2 9B It22
BenchmarkByScoreRank
Knowledge17
Math1
Writing1
Legacy · saturated or no longer reported3
No benchmark matches
Ranking?
Alternatives?
ModelScoreCostPer taskWhat changes
CheaperLlama 3.1 8B InstructunstatedOpensame-33%$0.02 / $0.04
BetterGPT-5 Minihighwins 71%+1,733%$0.053
OpenNVIDIA Nemotron 3.5 Lightning 30B A3B NVFP4wins 70%——Memory (4-bit): 4.6 GB → 11 GBCheaperLlama 3.2 3B InstructunstatedOpenwins 44%-47%$0.02 / $0.02BetterGPT-6 Lunamaxwins 73%+433%$0.068Function calling only with none · Chat Completions+3ModelScoreCostPer taskWhat changes
Llama 3.1 8B InstructunstatedOpensame-33%$0.02 / $0.04Llama 3.2 3B InstructunstatedOpenwins 44%-47%$0.02 / $0.02ModelScoreCostPer taskWhat changes
GPT-5 Minihighwins 71%+1,733%$0.053GPT-6 Lunamaxwins 73%+433%$0.068Function calling only with none · Chat Completions+3
Claude Haiku 5.5highwins 75%+433%$0.079Remove sampling parameters+5GPT-5.6 Lunaxhighwins 78%+1,100%$0.085GPT-5.6 Terraxhighwins 79%+11,900%$0.63Prompts with >272K input tokensModelScoreCostPer taskWhat changes
NVIDIA Nemotron 3.5 Lightning 30B A3B NVFP4Openwins 70%——Memory (4-bit): 4.6 GB → 11 GBGemma 3 12BunstatedOpenwins 68%+67%$0.05 / $0.10Memory (4-bit): 4.6 GB → 6.1 GB
Granite 4.2 8BunstatedOpenwins 66%+187%$0.06 / $0.25Granite 4.2 3BunstatedOpenwins 64%+40%$0.03 / $0.12Gemma 3n E4B InstructunstatedOpenwins 64%——ModelScoreCostPer taskWhat changes
Gemini 4 Argonhighwins 85%—*$1.62Synchronous output stops at 65,536+2Gemini 3.8 Flashhighwins 83%—*$3.49Sampling parameters are ignored+1Gemini 3.6 Flashhighwins 82%—*$1.60Gemini 3.1 Pro Previewunstatedwins 82%—$2 / $12Cannot use both thinking controlsGemini 3 Flash Previewunstatedwins 81%—$0.50 / $3Gemini 2.5 Prounstatedwins 80%—$1.25 / $10Gemma 4 31BunstatedOpenwins 79%—$0.10 / $0.34Memory (4-bit): 4.6 GB → 19 GBGemma 4 26B A4BunstatedOpenwins 78%—$0.15 / $0.60Memory (4-bit): 4.6 GB → 15 GBAll Google models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · compressed✓ 16 GB · 8-bit✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceGemma termscommercial use with conditions
Benchmarks
all categoriesChat & Knowledge
Arena Text · overall1,267#171
Arena Text · Chinese1,185#179
Arena Text · English1,219.7#183
Arena Text · French1,190.2#149
Arena Text · German1,186.1#146
Arena Text · instruction following1,178.1#184
Arena Text · Japanese1,144.3#131
Arena Text · Korean1,136.6#142
Arena Text · non-English1,187.7#181
Arena Text · Russian1,200.3#177
Arena Text · Spanish1,200.3#153
EQ-Bench 3636.2#51
MATH Level 521#30
Arena Text · overall · October 22, 20261,207.4#174
Arena Text · overall · October 15, 20261,207.4#165
GPQA Diamond · Epoch27.5#124
MMLU-Pro · official board52.1#15
OTIS Mock AIME 2024 20250.6#127
Writing & Language
Arena Text · creative writing1,259#155
EQ-Bench Creative Writing v3839.7#84
FAQ
How much does Gemma 2 9B It cost?›
$0.03 per 1M input tokens and $0.06 per 1M output tokens (cheapest host: DeepInfra).
Is Gemma 2 9B It free to use?›
There is no free API; the cheapest paid price is $0.03 per 1M input tokens. The weights are open, so you can run it yourself.
How good is Gemma 2 9B It, and what is it best at?›
#183 of 205 in Chat & Knowledge, #167 of 197 in Writing & Language.
Is Gemma 2 9B It better than Qwen 2 72B Instruct?›
On Arena Text · English, Gemma 2 9B It scores 1,219.7 vs 1,222 — about the same.
How can I use Gemma 2 9B It?›
Through the API of DeepInfra.
What is the context window of Gemma 2 9B It?›
8K tokens.
When was Gemma 2 9B It released? Is there a newer version?›
Released 2024-06-24. It is the current version in its line.
Is Gemma 2 9B It open source?›
Yes, the weights are open (gemma licence).
Can I run Gemma 2 9B It locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Gemma 2 9B It read images?›
No, it accepts text only.
What is Gemma 2 9B It?›
Gemma 2 9B It is an open-weights language model from Google, released Jun 2024.
What are the best alternatives to Gemma 2 9B It?›
Llama 3.1 8B Instruct (cheaper), GPT-5 Mini (better), NVIDIA Nemotron 3.5 Lightning 30B A3B NVFP4 (open weights).