50 models7 providersprices scraped nightly — no estimates
google · gemma familyVerified May 16, 2026

Gemma 2 9B IT

9B params · 8K context · gemma · released 2024-07 · 3 providers

Gemma 2 9B IT is Google DeepMind's mid-tier Gemma 2 model, a 9-billion-parameter instruction-tuned transformer released July 2024. It performs comparably to Llama 3.1 8B and Mistral 7B v0.3 on standard benchmarks, and Groq hosts it with some of the lowest latency numbers available for sub-10B models. Pricing across providers typically lands below $0.20 per million tokens, making it cost-competitive for high-throughput classification, extraction, and short-form generation. The 8K context window is the same bottleneck as the rest of the Gemma 2 line — document-QA or retrieval-augmented workloads that need 32K or more should look elsewhere. License is Gemma, permissive but not OSI-approved.

Cheapest right now
DeepInfra$0.0600/1M out
sticky while scrolling · verified May 16, 2026
Cheapest now
$0.0600
DeepInfra · fp16 · $/1M out
90-day price Δ
cheapest-host trend
Fastest TTFT
55 ms
Groq · 700 tok/s
Providers
3
indexed providers
✓ verified May 16, 2026
#Provider$/1M in$/1M outTTFT p50tok/sQuant
1DeepInfraCHEAPEST$0.0500$0.0600110 ms480fp16go ↗
2Fireworks AI$0.0800$0.0800130 ms450fp16go ↗
3GroqFASTEST$0.0800$0.080055 ms700fp8go ↗

Prices verified May 16, 2026 by scraping 3 providers. Methodology · Raw data