Claude Fable 5$22.000/MClaude Opus 4.8$11.000/MClaude Opus 4.7$11.000/MClaude Opus 4.6$11.000/MClaude Opus 4.5$33.000/MClaude Sonnet 3.7$6.600/MClaude Opus 3$33.000/MClaude 2.1$12.800/MClaude 2$12.800/MGPT-5.5$12.500/MGPT-5.2$5.425/MGPT-5.2-Codex$5.425/MGPT-5$3.875/MGPT-4.5$97.500/MGPT-4 Turbo Preview$16.000/MGPT-4$39.000/MGPT-4-32k$78.000/Mo3$19.000/Mo3-mini$2.090/Mo4-mini$2.090/Mo1$28.500/Mo1-mini$5.700/Mo1-preview$28.500/MGemini 3.5 Pro$5.000/MGemini 3.1 Pro$5.000/MGemini 3 Pro$5.000/MGemini 2.5 Pro$3.875/MGemini 1.5 Pro$2.375/MGemini 1.0 Ultra$12.000/MGemini 1.0 Pro$0.800/MClaude Fable 5$22.000/MClaude Opus 4.8$11.000/MClaude Opus 4.7$11.000/MClaude Opus 4.6$11.000/MClaude Opus 4.5$33.000/MClaude Sonnet 3.7$6.600/MClaude Opus 3$33.000/MClaude 2.1$12.800/MClaude 2$12.800/MGPT-5.5$12.500/MGPT-5.2$5.425/MGPT-5.2-Codex$5.425/MGPT-5$3.875/MGPT-4.5$97.500/MGPT-4 Turbo Preview$16.000/MGPT-4$39.000/MGPT-4-32k$78.000/Mo3$19.000/Mo3-mini$2.090/Mo4-mini$2.090/Mo1$28.500/Mo1-mini$5.700/Mo1-preview$28.500/MGemini 3.5 Pro$5.000/MGemini 3.1 Pro$5.000/MGemini 3 Pro$5.000/MGemini 2.5 Pro$3.875/MGemini 1.5 Pro$2.375/MGemini 1.0 Ultra$12.000/MGemini 1.0 Pro$0.800/M
BETA
Subscribe
← Back to Index·All IBM models
IBM
IBMEfficient

Granite 3.1 8B Instruct

Compact

IBM's open-weights Granite 3.1 8B. Apache-2.0 model with full IP indemnification when run on watsonx — the enterprise-procurement-friendly option.

Granite 3.1 8B Instruct is a efficient AI model from IBM. It costs $0.050 per million input tokens and $0.250 per million output tokens (blended $0.110/M), with a 128K-token context window.

Released Dec 2024Modalities textOfficial model page ↗API docs ↗Compare with another model →Estimate monthly cost →
INPUT
$0.050/M
per million input tokens
OUTPUT
$0.250/M
per million output tokens
BLENDED 70/30
$0.110/M
default reference rate · how it's calculated →
CONTEXT
128K
128,000 tokens
What it's good at
  • Apache 2.0 + IBM indemnity on watsonx
  • Long context (128K)
  • Single-GPU
Typical use cases
  • Enterprise on-prem chat
  • Regulated-industry RAG
  • Self-hosted production chat
Benchmarks
vs. best public score
MMLU65%
Multitask academic knowledge across 57 subjects.
HumanEval65%
Python function synthesis from docstrings.
Hand-curated from each provider's published reports and public leaderboards. Methodology varies across sources — treat as directional rather than authoritative.
More from IBM
See all 5 →
Granite 3.1 2B Instruct
Nano · 128K ctx
in $0.030/Mout $0.100/M
Granite 3.0 8B Dense
Previous Gen · 4K ctx
in $0.050/Mout $0.250/M
Granite Code 34B
Code · 8K ctx
in $0.350/Mout $1.050/M
Granite 20B Multilingual
Multilingual · 8K ctx
in $0.400/Mout $0.400/M
Frequently asked questions

How much does Granite 3.1 8B Instruct cost?

Granite 3.1 8B Instruct costs $0.050 per million input tokens and $0.250 per million output tokens, for a blended reference rate of $0.110 per million tokens.

What is Granite 3.1 8B Instruct's context window?

Granite 3.1 8B Instruct supports up to 128K tokens of context (128,000 tokens).

What is Granite 3.1 8B Instruct best for?

Granite 3.1 8B Instruct is well suited to Apache 2.0 + IBM indemnity on watsonx, Long context (128K) and Single-GPU.

Who makes Granite 3.1 8B Instruct?

Granite 3.1 8B Instruct is developed and served by IBM. It was released in Dec 2024.