Granite 3.1 8B Instruct

Released
2024-12-18
Last refreshed
2026-10-06
Status
Researched 181d ago
Open sourceCommercial use: permittedLong contextJSON / Tool use

Granite 3.1 8B Instruct is a released long context and json / tool use model with open-source and 128K context; evaluate it while provider pricing coverage matures.

IBM Research releases · 25 in the last 12 monthsChangelog →
Specifications
Family
Granite 3
Released
2024-12-18
Context
128K
Parameters
8B
Architecture
Decoder Only
Knowledge cutoff
2024-04
Openness
Open source
License
Apache 2.0OSI-approvedPermitted
Weights
Available
Code
Unknown
Created by
IBM Research

Creating reliable and adaptable AI solutions

Armonk, New York, United States
Founded 1945
Website
Pricing

No tracked provider pricing is available yet.

About

IBM Granite 3.1 8B Instruct is a significant update to Granite 3.0 8B with extended 128K context window (up from 4K in Granite 3.0). Dense decoder-only transformer with 40 layers, 4096 embedding, GQA (32 heads, 8 KV heads), RoPE. Supports 12 languages, code, tool-calling, RAG, and long-context document tasks. Open LLM Leaderboard avg 71.31. MMLU 65.34, GSM8K 73.84. Apache 2.0.

Providers

No active provider routes are tracked for this model yet.

Capabilities

No model capability flags are currently sourced.

Benchmark peer barsfor JSON / Tool use

BFCLRank 48 of 60
Claude Opus 4.5
77.5
Claude Sonnet 4.5
73.2
Qwen3.5-397B-A17B
72.9
Gemini 3 Pro
72.5
Granite 3.1 8B Instructcurrent
27.1

Benchmark scores(2)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.
BenchmarkScoreVersionEvaluationSource
BFCL
27.1v4Observed 2026-10-26
Configuration: Granite-3.1-8B-Instruct (FC)
Harness: BFCL V4 official evaluation (gorilla berkeley-function-call-leaderboard), FC mode
Evaluator: Gorilla LLM team (UC Berkeley)
Confidence: confirmed
Notes: BFCL V4 leaderboard (Last Updated 2026-04-12), entry 'Granite-3.1-8B-Instruct (FC)': rank 81, Overall Acc 27.10%; Multi Turn 7.50%, Web Search 0.50%, Memory 14.41%, Non-Live AST 78.33%, Live 60.33%, Irrelevance 79.98%. Retrieved 2026-10-26.
Source
Chatbot Arena
1209.0text_style_control overall 2026-10-22 (style control, default view)Observed 2026-10-26
Configuration: Granite 3.1 8B Instruct
Harness: LMArena blind pairwise human votes, Bradley-Terry
Evaluator: LMArena
Confidence: confirmed
Notes: LMArena text style control leaderboard, overall category, published 2026-10-22 (HF lmarena-ai/leaderboard-dataset @9761a3ce, latest split). Board entry 'Granite 3.1 8B Instruct': rank 346, Arena Score 1208.6 (95% CI 1197.5-1219.7), 3090 votes. Retrieved 2026-10-26.
Source

Source: LMArena (lmarena.ai), CC BY 4.0

Migration checks

No linked migration route is available for this model yet.

Compare Granite 3.1 8B Instruct with other models