Granite 3.3 8B (Non-reasoning) API Pricing

IBM · released 2025-04-16

Pricing

$0.03
input / 1M tokens
< 0.01¢ / 1K
$0.25
output / 1M tokens
0.025¢ / 1K
$0.085
blended / 1M tokens
< 0.01¢ / 1K

What it actually costs

WorkloadTokensCost
Short chat reply500 in / 300 out0.009¢
RAG query4,000 in / 500 out0.0245¢
Long document summary50,000 in / 1,000 out0.175¢
Code review of a 2k-line file30,000 in / 2,000 out0.14¢
1,000 chat replies/day for 30 days30,000 × (500 in / 300 out)$2.70

Cost = (input tokens ÷ 1,000,000 × $0.03) + (output tokens ÷ 1,000,000 × $0.25).

Benchmarks

4.9
Intelligence
#514 of 533
Coding
6.7
Math
#222 of 249
Tokens/sec
Time to first token

Similar intelligence, lower price

ModelIntelligenceBlended $/1M
Llama 3.1 Instruct 8B6.9$0.028
Sarvam 30B (high)6.6$0.047
Nova Micro5.9$0.061
Granite 4.1 8B6.6$0.063
NVIDIA Nemotron Nano 9B V2 (Reasoning)7.4$0.07

Leaderboard neighbours

ModelIntelligenceBlended $/1M
Molmo2-8B5.0
Qwen Chat 14B5.0
Gemma 3 27B Instruct4.9
Granite 3.3 8B (Non-reasoning)4.9$0.085
LFM2 8B A1B4.9
Qwen3 1.7B (Non-reasoning)4.9
Apertus 8B Instruct4.8$0.125

More from IBM

ModelIntelligenceBlended $/1M
Granite 4.2 30B14.8$0.282
Granite 4.2 8B11.1$0.107
Granite 4.2 3B9.1$0.052
Granite 4.1 30B7.4
Granite 4.1 8B6.6$0.063
Granite 4.0 H Small6.0$0.107
Granite 4.1 3B5.9
Granite 4.0 H 1B5.2

Count tokens & estimate cost for Granite 3.3 8B (Non-reasoning) See Granite 3.3 8B (Non-reasoning) on the leaderboard

Source: Artificial Analysis · last synced 2026-09-22

FAQ

Who develops Granite 3.3 8B (Non-reasoning)?

Granite 3.3 8B (Non-reasoning) is developed by IBM, released 2025-04-16.

How much does Granite 3.3 8B (Non-reasoning) cost per million tokens?

Granite 3.3 8B (Non-reasoning) costs $0.03 per 1M input tokens and $0.25 per 1M output tokens ($0.085 blended), per Artificial Analysis's official pricing data.

How does Granite 3.3 8B (Non-reasoning) rank on benchmarks?

Granite 3.3 8B (Non-reasoning) scores 4.9 on the Artificial Analysis Intelligence Index, ranking #514 of 533 models we track.

Is there a cheaper model with similar intelligence to Granite 3.3 8B (Non-reasoning)?

Yes — Llama 3.1 Instruct 8B scores similar intelligence (6.9) at $0.028/1M blended, versus $0.085/1M for Granite 3.3 8B (Non-reasoning).

Where does this data come from?

Pricing and benchmark data for Granite 3.3 8B (Non-reasoning) comes from Artificial Analysis (artificialanalysis.ai), synced daily. Last synced 2026-09-22.