Granite 4.2 3B granite-4.2-3b
ibm-granite · ibm-granite
#38 overall
- params
- 3.7B
- arch
- dense
- license
- apache-2.0 open weights
- released
- Aug 2026
- downloads/30d
- 16.7k
# architecture
- layers
- 40
- d_model
- 2560
- heads
- 40q / 8kv
- head dim
- 64
- vocab
- 100352
- family
- granite
signal path of one layer, generated from the registry's structured fields — dimension lines quote the model's real numbers; an MoE trace forks at the router, a dense trace runs straight through. Geometry fields come from the repo's config.json.
# why ranked
Overall: #38 — score 41.8 ● high — all signals present
| signal | weight | input (0–100) |
|---|---|---|
| aa_intelligence | 0.70 | 38.7 |
| bench_composite | 0.30 | 49.0 |
benchmark panel evidence:
| complete panel | score (0–100) |
|---|---|
| aime-2025-v1 | 49.0 |
Coding: provisional — score 15.8 ● low — a single signal; treat with caution
provisional: too few independent signals for a numbered position — the score is shown, but this model sorts after every ranked model.
| signal | weight | input (0–100) |
|---|---|---|
| aa_coding | 0.30 | 15.8 |
| aider_polyglot | 0.30 | — |
| swe_bench_verified | 0.40 | — |
missing signals are dropped and the remaining weights renormalized — never imputed.
# trend
# benchmarks
| benchmark | score | source | date |
|---|---|---|---|
| AA Coding Index | 17.5 | Artificial Analysis | — |
| AA Intelligence Index | 10.7 | Artificial Analysis | — |
| AIME 2025 | 0.8 / 1 | LLM Stats | — |
| Arena-hard-v2 | 0.3 / 1 | LLM Stats | — |
| Bfcl-v4 | 0.5 / 1 | LLM Stats | — |
# where to run
| provider | quant | ctx | $/M in | $/M out | $/M cache | price src | tps | uptime | ✓ |
|---|---|---|---|---|---|---|---|---|---|
| DeepInfra | 128k | $0.03 | $0.12 | — | deepinfra | 204 | — | ✓ |
sorted by blended price ((3·input + output) / 4 per 1M) · ✓ = the provider's own catalog confirms the offer · "via …" prices are what the aggregator routing the offer charges, not the provider's own list price
source aliases
- llmstats
granite-4.2-3b